{
  "schema": "programvla/benchmark-presentation/1",
  "variants": [
    {
      "id": "standalone-vlact-clean",
      "agent": "VLA",
      "label": "VLAct OFT · baseline",
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "provider": "StarVLA",
      "instruction_policy": "original_native",
      "instruction_source": "native",
      "mode": "image_only",
      "paired_campaign": true,
      "checkpoint_repo": "StarVLA/VLAct_Qwen3OFT_Robotwin_Finetune",
      "checkpoint_step": 50000,
      "evaluation_state": "complete",
      "running_episodes": 0,
      "queued_episodes": 0,
      "evaluation_date": "2026-11-15"
    },
    {
      "id": "pose-base-mode1",
      "agent": "VLA",
      "label": "progVLA-yz · Mode 1",
      "model": "progVLA-yz",
      "effort": "50k · native",
      "provider": "RLE-Bench",
      "instruction_policy": "original_native",
      "instruction_source": "native",
      "mode": "image_only",
      "paired_campaign": true,
      "evaluation_state": "complete",
      "running_episodes": 0,
      "queued_episodes": 0
    },
    {
      "id": "pose-base-mode2",
      "agent": "VLA",
      "label": "progVLA-yz · Mode 2",
      "model": "progVLA-yz",
      "effort": "50k · native",
      "provider": "RLE-Bench",
      "instruction_policy": "original_native",
      "instruction_source": "native",
      "mode": "reference_pose",
      "paired_campaign": true,
      "evaluation_state": "complete",
      "running_episodes": 0,
      "queued_episodes": 0
    },
    {
      "id": "jy-mode1",
      "agent": "VLA",
      "label": "progVLA-jy · Mode 1",
      "model": "progVLA-jy",
      "effort": "50k · native",
      "provider": "VLAct-jy",
      "instruction_policy": "original_native",
      "instruction_source": "native",
      "mode": "image_only",
      "paired_campaign": true,
      "evaluation_state": "complete",
      "running_episodes": 0,
      "queued_episodes": 0
    },
    {
      "id": "jy-mode2",
      "agent": "VLA",
      "label": "progVLA-jy · Mode 2",
      "model": "progVLA-jy",
      "effort": "50k · native",
      "provider": "VLAct-jy",
      "instruction_policy": "original_native",
      "instruction_source": "native",
      "mode": "reference_pose",
      "paired_campaign": true,
      "evaluation_state": "complete",
      "running_episodes": 0,
      "queued_episodes": 0
    },
    {
      "id": "kinex-tcp-posestate",
      "agent": "Kinex",
      "label": "Kinex + progVLA-jy · native instruction",
      "model": "GPT-6 Astra",
      "effort": "high",
      "provider": "OpenAI Codex",
      "vla_model": "progVLA-jy",
      "checkpoint_repo": "afhwoinavsnvo/VLAct-OFT-TCP-PoseState",
      "checkpoint_revision": "4bf022fbed59903ed398e23d5225f742e0c8956c",
      "checkpoint_filename": "final_model/pytorch_model.pt",
      "checkpoint": "research/kinex-pose-hard-native-20261113/checkpoint.json",
      "model_card": "research/kinex-pose-hard-native-20261113/PROTOCOL.md",
      "instruction_policy": "original_native",
      "instruction_source": "native",
      "agent_protocol": "optional-model-device",
      "input_contract": "images-current-target-tcp-xyzw-v1",
      "protocol_label": "Native unseen instruction · numeric TCP targets · progress off",
      "progress_head": false,
      "evaluation_date": "2026-11-16",
      "evaluation_status": "Tasks 01-05: onboard rerun 2026-11-15 · tasks 06-10: onboard rerun 2026-11-16 · tasks 11-50: 2026-11-16, 40 / 40 completed · 50 / 50 completed",
      "native_pose_campaign": true,
      "running_episodes": 0,
      "completed_valid_episodes": 50,
      "evaluation_state": "complete",
      "interrupted_episodes": 0,
      "frozen_pending_episodes": 0,
      "automatic_resume": false,
      "rerun_sources": {
        "tasks": [
          "task03/01",
          "task03/02",
          "task03/03",
          "task03/04",
          "task03/05"
        ],
        "protocol": "research/kinex-pose-hard-onboard-rerun-20261115/PROTOCOL.md",
        "report": "research/kinex-pose-hard-onboard-rerun-20261115/REPORT.md",
        "results": "research/kinex-pose-hard-onboard-rerun-20261115/summary.json",
        "source_identities": "research/kinex-pose-hard-onboard-rerun-20261115/source-identities.json",
        "validation": "research/kinex-pose-hard-onboard-rerun-20261115/validation.json"
      },
      "rerun_sources_06_10": {
        "tasks": [
          "task03/06",
          "task03/07",
          "task03/08",
          "task03/09",
          "task03/10"
        ],
        "protocol": "research/kinex-pose-hard-onboard-rerun-06-10-20261116/PROTOCOL.md",
        "report": "research/kinex-pose-hard-onboard-rerun-06-10-20261116/REPORT.md",
        "results": "research/kinex-pose-hard-onboard-rerun-06-10-20261116/summary.json",
        "source_identities": "research/kinex-pose-hard-onboard-rerun-06-10-20261116/source-identities.json",
        "validation": "research/kinex-pose-hard-onboard-rerun-06-10-20261116/validation.json"
      },
      "sources_11_50": {
        "tasks": [
          "task03/11",
          "task03/12",
          "task03/13",
          "task03/14",
          "task03/15",
          "task03/16",
          "task03/17",
          "task03/18",
          "task03/19",
          "task03/20",
          "task03/21",
          "task03/22",
          "task03/23",
          "task03/24",
          "task03/25",
          "task03/26",
          "task03/27",
          "task03/28",
          "task03/29",
          "task03/30",
          "task03/31",
          "task03/32",
          "task03/33",
          "task03/34",
          "task03/35",
          "task03/36",
          "task03/37",
          "task03/38",
          "task03/39",
          "task03/40",
          "task03/41",
          "task03/42",
          "task03/43",
          "task03/44",
          "task03/45",
          "task03/46",
          "task03/47",
          "task03/48",
          "task03/49",
          "task03/50"
        ],
        "protocol": "research/kinex-pose-hard-onboard-11-50-20261116/PROTOCOL.md",
        "report": "research/kinex-pose-hard-onboard-11-50-20261116/REPORT.md",
        "results": "research/kinex-pose-hard-onboard-11-50-20261116/summary.json",
        "source_identities": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "validation": "research/kinex-pose-hard-onboard-11-50-20261116/validation.json"
      }
    },
    {
      "id": "kinex-pose-base",
      "agent": "Kinex",
      "label": "Kinex + progVLA-yz · native instruction",
      "model": "GPT-6 Astra",
      "effort": "high",
      "provider": "OpenAI Codex",
      "vla_model": "progVLA-yz",
      "checkpoint_repo": "RLE-Bench/programvla-pose-head-ablations",
      "checkpoint_revision": "b6fc499a7b1d62b427cf05f6c8c77fc1ef3bba82",
      "checkpoint_filename": "pose_base/final_head.pt",
      "checkpoint": "research/kinex-pose-hard-native-20261113/checkpoint.json",
      "model_card": "research/kinex-pose-hard-native-20261113/PROTOCOL.md",
      "instruction_policy": "original_native",
      "instruction_source": "native",
      "agent_protocol": "optional-model-device",
      "input_contract": "images-current-target-tcp-xyzw-v1",
      "protocol_label": "Native unseen instruction · numeric TCP targets · progress off",
      "progress_head": false,
      "evaluation_date": "2026-11-13",
      "evaluation_status": "Stopped by user · 5 / 50 completed · 5 interrupted · 40 not run",
      "native_pose_campaign": true,
      "running_episodes": 0,
      "completed_valid_episodes": 5,
      "evaluation_state": "stopped_by_user",
      "interrupted_episodes": 5,
      "frozen_pending_episodes": 40,
      "automatic_resume": false
    },
    {
      "id": "kinex-pose-film",
      "agent": "Kinex",
      "label": "Kinex + progVLA-yz pose_film · native instruction",
      "model": "GPT-6 Astra",
      "effort": "high",
      "provider": "OpenAI Codex",
      "vla_model": "progVLA-yz pose_film",
      "checkpoint_repo": "RLE-Bench/programvla-pose-head-ablations",
      "checkpoint_revision": "b6fc499a7b1d62b427cf05f6c8c77fc1ef3bba82",
      "checkpoint_filename": "pose_film/final_head.pt",
      "checkpoint": "research/kinex-pose-film-hard-native-20261116/checkpoint.json",
      "model_card": "research/kinex-pose-film-hard-native-20261116/PROTOCOL.md",
      "instruction_policy": "original_native",
      "instruction_source": "native",
      "agent_protocol": "optional-model-device",
      "input_contract": "images-current-target-tcp-xyzw-v1",
      "protocol_label": "Native unseen instruction · numeric TCP targets · progress off",
      "progress_head": false,
      "evaluation_date": "2026-11-16",
      "evaluation_status": "2026-11-16 · 50 / 50 completed",
      "native_pose_campaign": true,
      "running_episodes": 0,
      "completed_valid_episodes": 50,
      "evaluation_state": "complete",
      "sources": {
        "tasks": [
          "task03/01",
          "task03/02",
          "task03/03",
          "task03/04",
          "task03/05",
          "task03/06",
          "task03/07",
          "task03/08",
          "task03/09",
          "task03/10",
          "task03/11",
          "task03/12",
          "task03/13",
          "task03/14",
          "task03/15",
          "task03/16",
          "task03/17",
          "task03/18",
          "task03/19",
          "task03/20",
          "task03/21",
          "task03/22",
          "task03/23",
          "task03/24",
          "task03/25",
          "task03/26",
          "task03/27",
          "task03/28",
          "task03/29",
          "task03/30",
          "task03/31",
          "task03/32",
          "task03/33",
          "task03/34",
          "task03/35",
          "task03/36",
          "task03/37",
          "task03/38",
          "task03/39",
          "task03/40",
          "task03/41",
          "task03/42",
          "task03/43",
          "task03/44",
          "task03/45",
          "task03/46",
          "task03/47",
          "task03/48",
          "task03/49",
          "task03/50"
        ],
        "protocol": "research/kinex-pose-film-hard-native-20261116/PROTOCOL.md",
        "report": "research/kinex-pose-film-hard-native-20261116/REPORT.md",
        "results": "research/kinex-pose-film-hard-native-20261116/summary.json",
        "source_identities": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "validation": "research/kinex-pose-film-hard-native-20261116/validation.json"
      }
    },
    {
      "id": "kinex-native",
      "agent": "Kinex",
      "label": "Kinex · native instruction",
      "model": "GPT-6 Astra",
      "effort": "high",
      "provider": "OpenAI Codex",
      "instruction_policy": "original_native",
      "instruction_source": "native",
      "evaluation_status": "50 / 50 completed valid; 0 running; 0 queued",
      "evaluation_date": "2026-11-13",
      "native_agent_campaign": true,
      "running_episodes": 0,
      "pending_episodes": 0,
      "evaluation_phase": "complete"
    },
    {
      "id": "kinex-vlact50k-native",
      "agent": "Kinex",
      "label": "Kinex + VLAct 50k · native instruction",
      "model": "GPT-6 Astra",
      "effort": "high",
      "provider": "OpenAI Codex + StarVLA",
      "instruction_policy": "original_native",
      "instruction_source": "native",
      "evaluation_status": "50 / 50 completed valid; 0 running; 0 queued",
      "evaluation_date": "2026-11-13",
      "native_agent_campaign": true,
      "vla_model": "VLAct Qwen3-VL-OFT",
      "checkpoint_repo": "StarVLA/VLAct_Qwen3OFT_Robotwin_Finetune",
      "checkpoint_step": 50000,
      "agent_protocol": "optional-model-device",
      "input_contract": "images-and-task-instruction",
      "checkpoint": "research/kinex-hard-native-20261113/checkpoint.json",
      "model_card": "research/kinex-hard-native-20261113/PROTOCOL.md",
      "running_episodes": 0,
      "pending_episodes": 0,
      "evaluation_phase": "complete"
    }
  ],
  "families": [
    {
      "id": "task03",
      "name": "RoboTwin 2.0 Hard",
      "tasks": 50
    }
  ],
  "tasks": [
    {
      "key": "task03/01",
      "slot": "01",
      "family": "task03",
      "native_id": "robotwin/handover-block-aloha-agilex-hard",
      "native_identity": {
        "allow_seed_substitution": false,
        "embodiment": "aloha_agilex",
        "max_steps": 800,
        "task_config": "demo_randomized",
        "task_name": "handover_block"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Handover block",
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100001,
        "pose-base-mode1": 100001,
        "pose-base-mode2": 100001,
        "jy-mode1": 100001,
        "jy-mode2": 100001,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 800,
          "max_control_steps": 800,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 800,
          "max_control_steps": 800,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
        "pose-base-mode1": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
        "pose-base-mode2": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
        "jy-mode1": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
        "jy-mode2": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/02",
      "slot": "02",
      "family": "task03",
      "native_id": "robotwin/place-dual-shoes-aloha-agilex-hard",
      "native_identity": {
        "allow_seed_substitution": false,
        "embodiment": "aloha_agilex",
        "max_steps": 600,
        "task_config": "demo_randomized",
        "task_name": "place_dual_shoes"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place dual shoes",
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100001,
        "pose-base-mode1": 100001,
        "pose-base-mode2": 100001,
        "jy-mode1": 100001,
        "jy-mode2": 100001,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
        "pose-base-mode1": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
        "pose-base-mode2": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
        "jy-mode1": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
        "jy-mode2": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/03",
      "slot": "03",
      "family": "task03",
      "native_id": "robotwin/put-bottles-dustbin-aloha-agilex-hard",
      "native_identity": {
        "allow_seed_substitution": false,
        "embodiment": "aloha_agilex",
        "max_steps": 1700,
        "task_config": "demo_randomized",
        "task_name": "put_bottles_dustbin"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Put bottles dustbin",
      "planned_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 1700,
        "max_control_steps": 1700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 2,
        "kinex-pose-base": 2,
        "kinex-vlact50k-native": 2,
        "kinex-native": 2,
        "kinex-pose-film": 2
      },
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1700,
          "max_control_steps": 1700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1700,
          "max_control_steps": 1700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1700,
          "max_control_steps": 1700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1700,
          "max_control_steps": 1700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1700,
          "max_control_steps": 1700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 2,
          "episodes": 1,
          "max_steps": 1700,
          "max_control_steps": 1700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 2,
          "episodes": 1,
          "max_steps": 1700,
          "max_control_steps": 1700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 2,
          "max_steps": 1700,
          "max_control_steps": 1700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 2,
          "max_steps": 1700,
          "max_control_steps": 1700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 2,
          "episodes": 1,
          "max_steps": 1700,
          "max_control_steps": 1700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
        "pose-base-mode1": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
        "pose-base-mode2": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
        "jy-mode1": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
        "jy-mode2": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/04",
      "slot": "04",
      "family": "task03",
      "native_id": "robotwin/scan-object-aloha-agilex-hard",
      "native_identity": {
        "task_name": "scan_object",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Scan object",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 500,
          "max_control_steps": 500,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 500,
          "max_control_steps": 500,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100001,
        "pose-base-mode1": 100001,
        "pose-base-mode2": 100001,
        "jy-mode1": 100001,
        "jy-mode2": 100001,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
        "pose-base-mode1": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
        "pose-base-mode2": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
        "jy-mode1": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
        "jy-mode2": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/05",
      "slot": "05",
      "family": "task03",
      "native_id": "robotwin/stack-bowls-three-aloha-agilex-hard",
      "native_identity": {
        "task_name": "stack_bowls_three",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Stack bowls three",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
        "pose-base-mode1": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
        "pose-base-mode2": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
        "jy-mode1": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
        "jy-mode2": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/06",
      "slot": "06",
      "family": "task03",
      "native_id": "robotwin/stack-blocks-three-aloha-agilex-hard",
      "native_identity": {
        "task_name": "stack_blocks_three",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Stack blocks three",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
        "pose-base-mode1": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
        "pose-base-mode2": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
        "jy-mode1": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
        "jy-mode2": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/07",
      "slot": "07",
      "family": "task03",
      "native_id": "robotwin/hanging-mug-aloha-agilex-hard",
      "native_identity": {
        "task_name": "hanging_mug",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Hanging mug",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 900,
          "max_control_steps": 900,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 900,
          "max_control_steps": 900,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
        "pose-base-mode1": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
        "pose-base-mode2": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
        "jy-mode1": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
        "jy-mode2": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/08",
      "slot": "08",
      "family": "task03",
      "native_id": "robotwin/put-object-cabinet-aloha-agilex-hard",
      "native_identity": {
        "task_name": "put_object_cabinet",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Put object cabinet",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100001,
        "pose-base-mode1": 100001,
        "pose-base-mode2": 100001,
        "jy-mode1": 100001,
        "jy-mode2": 100001,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
        "pose-base-mode1": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
        "pose-base-mode2": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
        "jy-mode1": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
        "jy-mode2": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/09",
      "slot": "09",
      "family": "task03",
      "native_id": "robotwin/blocks-ranking-size-aloha-agilex-hard",
      "native_identity": {
        "task_name": "blocks_ranking_size",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Blocks ranking size",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Place large block, medium block, and small block in order at the table's center.",
        "pose-base-mode1": "Place large block, medium block, and small block in order at the table's center.",
        "pose-base-mode2": "Place large block, medium block, and small block in order at the table's center.",
        "jy-mode1": "Place large block, medium block, and small block in order at the table's center.",
        "jy-mode2": "Place large block, medium block, and small block in order at the table's center.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/10",
      "slot": "10",
      "family": "task03",
      "native_id": "robotwin/lift-pot-aloha-agilex-hard",
      "native_identity": {
        "task_name": "lift_pot",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Lift pot · coordinated bimanual lifting",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Lift the dark gray cooking kitchenpot using both arms",
        "pose-base-mode1": "Lift the dark gray cooking kitchenpot using both arms",
        "pose-base-mode2": "Lift the dark gray cooking kitchenpot using both arms",
        "jy-mode1": "Lift the dark gray cooking kitchenpot using both arms",
        "jy-mode2": "Lift the dark gray cooking kitchenpot using both arms",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/11",
      "slot": "11",
      "family": "task03",
      "native_id": "robotwin/adjust-bottle-aloha-agilex-hard",
      "native_identity": {
        "task_name": "adjust_bottle",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Adjust bottle",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 2,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 2,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 2,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 2,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 2,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 2,
        "kinex-pose-base": 2,
        "kinex-vlact50k-native": 2,
        "kinex-native": 2,
        "kinex-pose-film": 2
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use the left arm to grab the brown bottle upright",
        "pose-base-mode1": "Use the left arm to grab the brown bottle upright",
        "pose-base-mode2": "Use the left arm to grab the brown bottle upright",
        "jy-mode1": "Use the left arm to grab the brown bottle upright",
        "jy-mode2": "Use the left arm to grab the brown bottle upright",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/12",
      "slot": "12",
      "family": "task03",
      "native_id": "robotwin/beat-block-hammer-aloha-agilex-hard",
      "native_identity": {
        "task_name": "beat_block_hammer",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Beat block hammer",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use the left arm to grab the hammer with black handle then beat block",
        "pose-base-mode1": "Use the left arm to grab the hammer with black handle then beat block",
        "pose-base-mode2": "Use the left arm to grab the hammer with black handle then beat block",
        "jy-mode1": "Use the left arm to grab the hammer with black handle then beat block",
        "jy-mode2": "Use the left arm to grab the hammer with black handle then beat block",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/13",
      "slot": "13",
      "family": "task03",
      "native_id": "robotwin/blocks-ranking-rgb-aloha-agilex-hard",
      "native_identity": {
        "task_name": "blocks_ranking_rgb",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Blocks ranking rgb",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 1200,
          "max_control_steps": 1200,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
        "pose-base-mode1": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
        "pose-base-mode2": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
        "jy-mode1": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
        "jy-mode2": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/14",
      "slot": "14",
      "family": "task03",
      "native_id": "robotwin/click-alarmclock-aloha-agilex-hard",
      "native_identity": {
        "task_name": "click_alarmclock",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Click alarmclock",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Tap the center top button of the alarm-clock with beige screen",
        "pose-base-mode1": "Tap the center top button of the alarm-clock with beige screen",
        "pose-base-mode2": "Tap the center top button of the alarm-clock with beige screen",
        "jy-mode1": "Tap the center top button of the alarm-clock with beige screen",
        "jy-mode2": "Tap the center top button of the alarm-clock with beige screen",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/15",
      "slot": "15",
      "family": "task03",
      "native_id": "robotwin/click-bell-aloha-agilex-hard",
      "native_identity": {
        "task_name": "click_bell",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Click bell",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use the left arm to press the medium-sized bell's top center.",
        "pose-base-mode1": "Use the left arm to press the medium-sized bell's top center.",
        "pose-base-mode2": "Use the left arm to press the medium-sized bell's top center.",
        "jy-mode1": "Use the left arm to press the medium-sized bell's top center.",
        "jy-mode2": "Use the left arm to press the medium-sized bell's top center.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/16",
      "slot": "16",
      "family": "task03",
      "native_id": "robotwin/dump-bin-bigbin-aloha-agilex-hard",
      "native_identity": {
        "task_name": "dump_bin_bigbin",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Dump bin bigbin",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100006,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100006,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100006,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100006,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100006,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100006,
        "pose-base-mode1": 100006,
        "pose-base-mode2": 100006,
        "jy-mode1": 100006,
        "jy-mode2": 100006,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
        "pose-base-mode1": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
        "pose-base-mode2": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
        "jy-mode1": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
        "jy-mode2": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/17",
      "slot": "17",
      "family": "task03",
      "native_id": "robotwin/grab-roller-aloha-agilex-hard",
      "native_identity": {
        "task_name": "grab_roller",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Grab roller",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use arms to grab the smooth wooden roller.",
        "pose-base-mode1": "Use arms to grab the smooth wooden roller.",
        "pose-base-mode2": "Use arms to grab the smooth wooden roller.",
        "jy-mode1": "Use arms to grab the smooth wooden roller.",
        "jy-mode2": "Use arms to grab the smooth wooden roller.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/18",
      "slot": "18",
      "family": "task03",
      "native_id": "robotwin/handover-mic-aloha-agilex-hard",
      "native_identity": {
        "task_name": "handover_mic",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Handover mic",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 600,
          "max_control_steps": 600,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 600,
          "max_control_steps": 600,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 600,
          "max_control_steps": 600,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use one arm to hold the microphone used for audio input and hand it over.",
        "pose-base-mode1": "Use one arm to hold the microphone used for audio input and hand it over.",
        "pose-base-mode2": "Use one arm to hold the microphone used for audio input and hand it over.",
        "jy-mode1": "Use one arm to hold the microphone used for audio input and hand it over.",
        "jy-mode2": "Use one arm to hold the microphone used for audio input and hand it over.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/19",
      "slot": "19",
      "family": "task03",
      "native_id": "robotwin/move-can-pot-aloha-agilex-hard",
      "native_identity": {
        "task_name": "move_can_pot",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Move can pot",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
        "pose-base-mode1": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
        "pose-base-mode2": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
        "jy-mode1": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
        "jy-mode2": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/20",
      "slot": "20",
      "family": "task03",
      "native_id": "robotwin/move-pillbottle-pad-aloha-agilex-hard",
      "native_identity": {
        "task_name": "move_pillbottle_pad",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Move pillbottle pad",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100001,
        "pose-base-mode1": 100001,
        "pose-base-mode2": 100001,
        "jy-mode1": 100001,
        "jy-mode2": 100001,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Grab the white cylindrical bottle and set it on the pad",
        "pose-base-mode1": "Grab the white cylindrical bottle and set it on the pad",
        "pose-base-mode2": "Grab the white cylindrical bottle and set it on the pad",
        "jy-mode1": "Grab the white cylindrical bottle and set it on the pad",
        "jy-mode2": "Grab the white cylindrical bottle and set it on the pad",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/21",
      "slot": "21",
      "family": "task03",
      "native_id": "robotwin/move-playingcard-away-aloha-agilex-hard",
      "native_identity": {
        "task_name": "move_playingcard_away",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Move playingcard away",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use the right arm to grab the hard black box for card deck and relocate it.",
        "pose-base-mode1": "Use the right arm to grab the hard black box for card deck and relocate it.",
        "pose-base-mode2": "Use the right arm to grab the hard black box for card deck and relocate it.",
        "jy-mode1": "Use the right arm to grab the hard black box for card deck and relocate it.",
        "jy-mode2": "Use the right arm to grab the hard black box for card deck and relocate it.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/22",
      "slot": "22",
      "family": "task03",
      "native_id": "robotwin/move-stapler-pad-aloha-agilex-hard",
      "native_identity": {
        "task_name": "move_stapler_pad",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Move stapler pad",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use the left arm to move the plastic stapler to Yellow mat",
        "pose-base-mode1": "Use the left arm to move the plastic stapler to Yellow mat",
        "pose-base-mode2": "Use the left arm to move the plastic stapler to Yellow mat",
        "jy-mode1": "Use the left arm to move the plastic stapler to Yellow mat",
        "jy-mode2": "Use the left arm to move the plastic stapler to Yellow mat",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/23",
      "slot": "23",
      "family": "task03",
      "native_id": "robotwin/open-laptop-aloha-agilex-hard",
      "native_identity": {
        "task_name": "open_laptop",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Open laptop",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Open the flat orange laptop with one motion.",
        "pose-base-mode1": "Open the flat orange laptop with one motion.",
        "pose-base-mode2": "Open the flat orange laptop with one motion.",
        "jy-mode1": "Open the flat orange laptop with one motion.",
        "jy-mode2": "Open the flat orange laptop with one motion.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/24",
      "slot": "24",
      "family": "task03",
      "native_id": "robotwin/open-microwave-aloha-agilex-hard",
      "native_identity": {
        "task_name": "open_microwave",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Open microwave",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1500,
          "max_control_steps": 1500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1500,
          "max_control_steps": 1500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1500,
          "max_control_steps": 1500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1500,
          "max_control_steps": 1500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 1500,
          "max_control_steps": 1500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 1500,
          "max_control_steps": 1500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 1500,
          "max_control_steps": 1500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 1500,
          "max_control_steps": 1500,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 1500,
          "max_control_steps": 1500,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 1500,
          "max_control_steps": 1500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1500,
        "max_control_steps": 1500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
        "pose-base-mode1": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
        "pose-base-mode2": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
        "jy-mode1": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
        "jy-mode2": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/25",
      "slot": "25",
      "family": "task03",
      "native_id": "robotwin/pick-diverse-bottles-aloha-agilex-hard",
      "native_identity": {
        "task_name": "pick_diverse_bottles",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Pick diverse bottles",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100005,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100005,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100005,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100005,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100005,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 2,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 2,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 2,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 2,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 2,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100005,
        "pose-base-mode1": 100005,
        "pose-base-mode2": 100005,
        "jy-mode1": 100005,
        "jy-mode2": 100005,
        "kinex-tcp-posestate": 2,
        "kinex-pose-base": 2,
        "kinex-vlact50k-native": 2,
        "kinex-native": 2,
        "kinex-pose-film": 2
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
        "pose-base-mode1": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
        "pose-base-mode2": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
        "jy-mode1": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
        "jy-mode2": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/26",
      "slot": "26",
      "family": "task03",
      "native_id": "robotwin/pick-dual-bottles-aloha-agilex-hard",
      "native_identity": {
        "task_name": "pick_dual_bottles",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Pick dual bottles",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
        "pose-base-mode1": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
        "pose-base-mode2": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
        "jy-mode1": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
        "jy-mode2": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/27",
      "slot": "27",
      "family": "task03",
      "native_id": "robotwin/place-a2b-left-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_a2b_left",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place a2b left",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
        "pose-base-mode1": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
        "pose-base-mode2": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
        "jy-mode1": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
        "jy-mode2": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/28",
      "slot": "28",
      "family": "task03",
      "native_id": "robotwin/place-a2b-right-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_a2b_right",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place a2b right",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
        "pose-base-mode1": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
        "pose-base-mode2": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
        "jy-mode1": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
        "jy-mode2": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/29",
      "slot": "29",
      "family": "task03",
      "native_id": "robotwin/place-bread-basket-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_bread_basket",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place bread basket",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
        "pose-base-mode1": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
        "pose-base-mode2": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
        "jy-mode1": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
        "jy-mode2": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/30",
      "slot": "30",
      "family": "task03",
      "native_id": "robotwin/place-bread-skillet-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_bread_skillet",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place bread skillet",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 500,
          "max_control_steps": 500,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 500,
          "max_control_steps": 500,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100001,
        "pose-base-mode1": 100001,
        "pose-base-mode2": 100001,
        "jy-mode1": 100001,
        "jy-mode2": 100001,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
        "pose-base-mode1": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
        "pose-base-mode2": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
        "jy-mode1": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
        "jy-mode2": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/31",
      "slot": "31",
      "family": "task03",
      "native_id": "robotwin/place-burger-fries-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_burger_fries",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place burger fries",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 500,
          "max_control_steps": 500,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 500,
          "max_control_steps": 500,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
        "pose-base-mode1": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
        "pose-base-mode2": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
        "jy-mode1": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
        "jy-mode2": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/32",
      "slot": "32",
      "family": "task03",
      "native_id": "robotwin/place-can-basket-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_can_basket",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place can basket",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
        "pose-base-mode1": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
        "pose-base-mode2": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
        "jy-mode1": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
        "jy-mode2": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/33",
      "slot": "33",
      "family": "task03",
      "native_id": "robotwin/place-cans-plasticbox-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_cans_plasticbox",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place cans plasticbox",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
        "pose-base-mode1": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
        "pose-base-mode2": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
        "jy-mode1": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
        "jy-mode2": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/34",
      "slot": "34",
      "family": "task03",
      "native_id": "robotwin/place-container-plate-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_container_plate",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place container plate",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Place the curved bowl with smooth texture on the gray plate for dining.",
        "pose-base-mode1": "Place the curved bowl with smooth texture on the gray plate for dining.",
        "pose-base-mode2": "Place the curved bowl with smooth texture on the gray plate for dining.",
        "jy-mode1": "Place the curved bowl with smooth texture on the gray plate for dining.",
        "jy-mode2": "Place the curved bowl with smooth texture on the gray plate for dining.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/35",
      "slot": "35",
      "family": "task03",
      "native_id": "robotwin/place-empty-cup-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_empty_cup",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place empty cup",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 500,
          "max_control_steps": 500,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 500,
          "max_control_steps": 500,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
        "pose-base-mode1": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
        "pose-base-mode2": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
        "jy-mode1": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
        "jy-mode2": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/36",
      "slot": "36",
      "family": "task03",
      "native_id": "robotwin/place-fan-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_fan",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place fan",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
        "pose-base-mode1": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
        "pose-base-mode2": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
        "jy-mode1": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
        "jy-mode2": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/37",
      "slot": "37",
      "family": "task03",
      "native_id": "robotwin/place-mouse-pad-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_mouse_pad",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place mouse pad",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 2,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 2,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 2,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 2,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 2,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 2,
        "kinex-pose-base": 2,
        "kinex-vlact50k-native": 2,
        "kinex-native": 2,
        "kinex-pose-film": 2
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
        "pose-base-mode1": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
        "pose-base-mode2": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
        "jy-mode1": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
        "jy-mode2": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/38",
      "slot": "38",
      "family": "task03",
      "native_id": "robotwin/place-object-basket-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_object_basket",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place object basket",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
        "pose-base-mode1": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
        "pose-base-mode2": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
        "jy-mode1": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
        "jy-mode2": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/39",
      "slot": "39",
      "family": "task03",
      "native_id": "robotwin/place-object-scale-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_object_scale",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place object scale",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
        "pose-base-mode1": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
        "pose-base-mode2": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
        "jy-mode1": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
        "jy-mode2": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/40",
      "slot": "40",
      "family": "task03",
      "native_id": "robotwin/place-object-stand-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_object_stand",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place object stand",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100001,
        "pose-base-mode1": 100001,
        "pose-base-mode2": 100001,
        "jy-mode1": 100001,
        "jy-mode2": 100001,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Place the white stapler on the compact black hollow displaystand with the left arm.",
        "pose-base-mode1": "Place the white stapler on the compact black hollow displaystand with the left arm.",
        "pose-base-mode2": "Place the white stapler on the compact black hollow displaystand with the left arm.",
        "jy-mode1": "Place the white stapler on the compact black hollow displaystand with the left arm.",
        "jy-mode2": "Place the white stapler on the compact black hollow displaystand with the left arm.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/41",
      "slot": "41",
      "family": "task03",
      "native_id": "robotwin/place-phone-stand-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_phone_stand",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place phone stand",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Set the phone with flat black screen on the angled brown phonestand.",
        "pose-base-mode1": "Set the phone with flat black screen on the angled brown phonestand.",
        "pose-base-mode2": "Set the phone with flat black screen on the angled brown phonestand.",
        "jy-mode1": "Set the phone with flat black screen on the angled brown phonestand.",
        "jy-mode2": "Set the phone with flat black screen on the angled brown phonestand.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/42",
      "slot": "42",
      "family": "task03",
      "native_id": "robotwin/place-shoe-aloha-agilex-hard",
      "native_identity": {
        "task_name": "place_shoe",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Place shoe",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 500,
          "max_control_steps": 500,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100001,
        "pose-base-mode1": 100001,
        "pose-base-mode2": 100001,
        "jy-mode1": 100001,
        "jy-mode2": 100001,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Pick up the medium adult shoe from the table and set it on the mat",
        "pose-base-mode1": "Pick up the medium adult shoe from the table and set it on the mat",
        "pose-base-mode2": "Pick up the medium adult shoe from the table and set it on the mat",
        "jy-mode1": "Pick up the medium adult shoe from the table and set it on the mat",
        "jy-mode2": "Pick up the medium adult shoe from the table and set it on the mat",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/43",
      "slot": "43",
      "family": "task03",
      "native_id": "robotwin/press-stapler-aloha-agilex-hard",
      "native_identity": {
        "task_name": "press_stapler",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Press stapler",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Press the plastic stapler firmly using your arm.",
        "pose-base-mode1": "Press the plastic stapler firmly using your arm.",
        "pose-base-mode2": "Press the plastic stapler firmly using your arm.",
        "jy-mode1": "Press the plastic stapler firmly using your arm.",
        "jy-mode2": "Press the plastic stapler firmly using your arm.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/44",
      "slot": "44",
      "family": "task03",
      "native_id": "robotwin/rotate-qrcode-aloha-agilex-hard",
      "native_identity": {
        "task_name": "rotate_qrcode",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Rotate qrcode",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
        "pose-base-mode1": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
        "pose-base-mode2": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
        "jy-mode1": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
        "jy-mode2": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/45",
      "slot": "45",
      "family": "task03",
      "native_id": "robotwin/shake-bottle-aloha-agilex-hard",
      "native_identity": {
        "task_name": "shake_bottle",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Shake bottle",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100001,
        "pose-base-mode1": 100001,
        "pose-base-mode2": 100001,
        "jy-mode1": 100001,
        "jy-mode2": 100001,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Pick up the smooth bottle with blue text and shake it.",
        "pose-base-mode1": "Pick up the smooth bottle with blue text and shake it.",
        "pose-base-mode2": "Pick up the smooth bottle with blue text and shake it.",
        "jy-mode1": "Pick up the smooth bottle with blue text and shake it.",
        "jy-mode2": "Pick up the smooth bottle with blue text and shake it.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/46",
      "slot": "46",
      "family": "task03",
      "native_id": "robotwin/shake-bottle-horizontally-aloha-agilex-hard",
      "native_identity": {
        "task_name": "shake_bottle_horizontally",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Shake bottle horizontally",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100001,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 700,
          "max_control_steps": 700,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 700,
          "max_control_steps": 700,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100001,
        "pose-base-mode1": 100001,
        "pose-base-mode2": 100001,
        "jy-mode1": 100001,
        "jy-mode2": 100001,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
        "pose-base-mode1": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
        "pose-base-mode2": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
        "jy-mode1": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
        "jy-mode2": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/47",
      "slot": "47",
      "family": "task03",
      "native_id": "robotwin/stack-blocks-two-aloha-agilex-hard",
      "native_identity": {
        "task_name": "stack_blocks_two",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Stack blocks two",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 800,
          "max_control_steps": 800,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 800,
          "max_control_steps": 800,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 800,
          "max_control_steps": 800,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Move red block to the center, then put green block on top.",
        "pose-base-mode1": "Move red block to the center, then put green block on top.",
        "pose-base-mode2": "Move red block to the center, then put green block on top.",
        "jy-mode1": "Move red block to the center, then put green block on top.",
        "jy-mode2": "Move red block to the center, then put green block on top.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/48",
      "slot": "48",
      "family": "task03",
      "native_id": "robotwin/stack-bowls-two-aloha-agilex-hard",
      "native_identity": {
        "task_name": "stack_bowls_two",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Stack bowls two",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 1,
          "episodes": 1,
          "max_steps": 900,
          "max_control_steps": 900,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 1,
        "kinex-pose-base": 1,
        "kinex-vlact50k-native": 1,
        "kinex-native": 1,
        "kinex-pose-film": 1
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
        "pose-base-mode1": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
        "pose-base-mode2": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
        "jy-mode1": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
        "jy-mode2": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/49",
      "slot": "49",
      "family": "task03",
      "native_id": "robotwin/stamp-seal-aloha-agilex-hard",
      "native_identity": {
        "task_name": "stamp_seal",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Stamp seal",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100003,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100003,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100003,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100003,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100003,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100003,
        "pose-base-mode1": 100003,
        "pose-base-mode2": 100003,
        "jy-mode1": 100003,
        "jy-mode2": 100003,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Grab the block-shaped seal with angled edges, press onto Blue",
        "pose-base-mode1": "Grab the block-shaped seal with angled edges, press onto Blue",
        "pose-base-mode2": "Grab the block-shaped seal with angled edges, press onto Blue",
        "jy-mode1": "Grab the block-shaped seal with angled edges, press onto Blue",
        "jy-mode2": "Grab the block-shaped seal with angled edges, press onto Blue",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    },
    {
      "key": "task03/50",
      "slot": "50",
      "family": "task03",
      "native_id": "robotwin/turn-switch-aloha-agilex-hard",
      "native_identity": {
        "task_name": "turn_switch",
        "task_config": "demo_randomized",
        "embodiment": "aloha_agilex"
      },
      "native_instruction": null,
      "catalog_instruction": "Native unseen instruction; see the selected episode.",
      "title": "Turn switch",
      "planned_variants": [
        "standalone-vlact-clean",
        "pose-base-mode1",
        "pose-base-mode2",
        "jy-mode1",
        "jy-mode2",
        "kinex-tcp-posestate",
        "kinex-pose-base",
        "kinex-pose-film",
        "kinex-native",
        "kinex-vlact50k-native"
      ],
      "planned_protocol_by_variant": {
        "standalone-vlact-clean": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "pose-base-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode1": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "image_only",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "jy-mode2": {
          "seed": 100000,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_type": "unseen",
          "provider": "robotwin-official",
          "control_frequency_hz": null,
          "mode": "native_trajectory",
          "action_chunk_size": 32,
          "model_seed": 42,
          "pose_conditioning": "reference_pose",
          "progress_head": false,
          "admission": "Frozen successful native reference admission; exact initial-state verification."
        },
        "kinex-tcp-posestate": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-model-device-onboard-rerun-20261115"
        },
        "kinex-pose-base": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06
        },
        "kinex-vlact50k-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-native": {
          "seed": 0,
          "max_steps": 400,
          "max_control_steps": 400,
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "episodes": 1,
          "provider": "rosbridge",
          "task_config": "demo_randomized"
        },
        "kinex-pose-film": {
          "seed": 0,
          "episodes": 1,
          "max_steps": 400,
          "max_control_steps": 400,
          "task_config": "demo_randomized",
          "instruction_source": "native",
          "instruction_policy": "original_native",
          "instruction_type": "unseen",
          "provider": "rosbridge",
          "control_frequency_hz": 25,
          "mode": "stepped",
          "admission": "Unchanged authored scene seed, no expert-success filtering, no valid-outcome retry.",
          "agent_protocol": "optional-model-device",
          "input_contract": "images-current-target-tcp-xyzw-v1",
          "progress_head": false,
          "ticks": 4,
          "tick_s": 0.04,
          "source_action_period_s": 0.06,
          "runtime_revision": "pose-film-model-device-20261116"
        }
      },
      "planned_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "planned_seed_by_variant": {
        "standalone-vlact-clean": 100000,
        "pose-base-mode1": 100000,
        "pose-base-mode2": 100000,
        "jy-mode1": 100000,
        "jy-mode2": 100000,
        "kinex-tcp-posestate": 0,
        "kinex-pose-base": 0,
        "kinex-vlact50k-native": 0,
        "kinex-native": 0,
        "kinex-pose-film": 0
      },
      "catalog_instruction_by_variant": {
        "standalone-vlact-clean": "Press the flat square gray switch using the left arm",
        "pose-base-mode1": "Press the flat square gray switch using the left arm",
        "pose-base-mode2": "Press the flat square gray switch using the left arm",
        "jy-mode1": "Press the flat square gray switch using the left arm",
        "jy-mode2": "Press the flat square gray switch using the left arm",
        "kinex-tcp-posestate": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-base": "Original native unseen instruction resolved from the initialized RoboTwin scene.",
        "kinex-pose-film": "Original native unseen instruction resolved from the initialized RoboTwin scene."
      }
    }
  ],
  "title": "RoboTwin 2.0 Hard · Native instruction",
  "updated_at": "2026-11-16T15:49:32.802420+00:00",
  "benchmark_complete": false,
  "complete": false,
  "episodes": [
    {
      "id": "hard-native-baseline-image_only-01-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/01",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "01",
      "seed": 100001,
      "source_episode_id": "01-handover-block-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 800
      },
      "steps": 800,
      "simulation_time_s": 205.200009746477,
      "wall_time_s": 1973.23179829,
      "timing": {
        "wall_time_s": 1973.23179829,
        "simulation_time_s": 205.200009746477,
        "physics_steps": 51300,
        "wall_time_start_ns": 1794764147234279477,
        "wall_time_end_ns": 1794766120466080463,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
      "native_instruction": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 800 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-01-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 25,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 20.0,
        "native_frames": 800,
        "published_frames": 800,
        "decode_ok": true,
        "sha256": "f8d41f16f55caa80aab4826f510b65b32731d23b69598cfabd5afdc949eefa15",
        "source_sha256": "9a71b8523f6423f5272429a7dd159712867ded76d16ffe4d9add07379c9d46c8",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-02-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/02",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "02",
      "seed": 100001,
      "source_episode_id": "02-place-dual-shoes-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 600
      },
      "steps": 600,
      "simulation_time_s": 208.56000990606844,
      "wall_time_s": 1104.450053491,
      "timing": {
        "wall_time_s": 1104.450053491,
        "simulation_time_s": 208.56000990606844,
        "physics_steps": 52140,
        "wall_time_start_ns": 1794764151034786247,
        "wall_time_end_ns": 1794765255484844016,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
      "native_instruction": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 600 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-02-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 15.0,
        "native_frames": 600,
        "published_frames": 600,
        "decode_ok": true,
        "sha256": "2504812333a2375252709f5af6a6149c4de07a353e0b9bbf3abe3649ccbdf478",
        "source_sha256": "684a3084830c3b3de94c6d87a9b793e118b249ccd231cb71625bf36e3190ed44",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-03-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/03",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "03",
      "seed": 100000,
      "source_episode_id": "03-put-bottles-dustbin-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1700
      },
      "steps": 1700,
      "simulation_time_s": 580.1720275566913,
      "wall_time_s": 2411.949171095,
      "timing": {
        "wall_time_s": 2411.949171095,
        "simulation_time_s": 580.1720275566913,
        "physics_steps": 145043,
        "wall_time_start_ns": 1794764152282335388,
        "wall_time_end_ns": 1794766564231509582,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
      "native_instruction": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-03-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1700,
        "max_control_steps": 1700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 54,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 42.5,
        "native_frames": 1700,
        "published_frames": 1700,
        "decode_ok": true,
        "sha256": "803730e101c001780ee092e4b0fc9f57319b7dd7e60a55763a4c13cb87aaac3a",
        "source_sha256": "a9a908354a73d3f30847afe2e5a935d21a55fe12a6a9c0d8a855d3f37898fd83",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184,
          1216,
          1248,
          1280,
          1312,
          1344,
          1376,
          1408,
          1440,
          1472,
          1504,
          1536,
          1568,
          1600,
          1632,
          1664,
          1696
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-04-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/04",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "04",
      "seed": 100001,
      "source_episode_id": "04-scan-object-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 161
      },
      "steps": 161,
      "simulation_time_s": 58.396002773661166,
      "wall_time_s": 359.85189665,
      "timing": {
        "wall_time_s": 359.85189665,
        "simulation_time_s": 58.396002773661166,
        "physics_steps": 14599,
        "wall_time_start_ns": 1794605908986665293,
        "wall_time_end_ns": 1794606268838572022,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
      "native_instruction": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 161 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-04-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 6,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 4.05,
        "native_frames": 162,
        "published_frames": 162,
        "decode_ok": true,
        "sha256": "b0e85cbd1e8f1a245c983d7d98c947dda1009fcaea02a974aed9de389af41c1f",
        "source_sha256": "ef66e01d5bc20aeb9a2875d2fb77d50b0d769159f5f63c334701ecc1016af527",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-05-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/05",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "05",
      "seed": 100000,
      "source_episode_id": "05-stack-bowls-three-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 602
      },
      "steps": 602,
      "simulation_time_s": 173.70400825049728,
      "wall_time_s": 1330.906125691,
      "timing": {
        "wall_time_s": 1330.906125691,
        "simulation_time_s": 173.70400825049728,
        "physics_steps": 43426,
        "wall_time_start_ns": 1794764161252141290,
        "wall_time_end_ns": 1794765492158278127,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
      "native_instruction": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 602 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-05-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 15.075,
        "native_frames": 603,
        "published_frames": 603,
        "decode_ok": true,
        "sha256": "3318ab0e40f8603a494c1b052808fe89bdd1d5d86b57c56948fa0018037837d5",
        "source_sha256": "30a31253b3a63b4ae4ac8d660e50a9da8666696ebc33fded00eaf08f87eb807c",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-06-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/06",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "06",
      "seed": 100000,
      "source_episode_id": "06-stack-blocks-three-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 325.3840154549107,
      "wall_time_s": 1464.987224307,
      "timing": {
        "wall_time_s": 1464.987224307,
        "simulation_time_s": 325.3840154549107,
        "physics_steps": 81346,
        "wall_time_start_ns": 1794764193361990188,
        "wall_time_end_ns": 1794765658349217448,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
      "native_instruction": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-06-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "eff50d719e77351ed7d8db0dc9bb1024fb9cd2859d3a2006846b3c0fa0256f3a",
        "source_sha256": "893249878dc400fbb393dd4535fde74557528c21bc957baca0ac110281a66c28",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-07-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/07",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "07",
      "seed": 100000,
      "source_episode_id": "07-hanging-mug-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 900
      },
      "steps": 900,
      "simulation_time_s": 200.27200951240957,
      "wall_time_s": 1386.024562243,
      "timing": {
        "wall_time_s": 1386.024562243,
        "simulation_time_s": 200.27200951240957,
        "physics_steps": 50068,
        "wall_time_start_ns": 1794764187163008233,
        "wall_time_end_ns": 1794765573187574346,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
      "native_instruction": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 900 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-07-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 29,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 22.5,
        "native_frames": 900,
        "published_frames": 900,
        "decode_ok": true,
        "sha256": "66d679cb264e638213836a731264e346cebb7cf6dbbba4511d25e1c61c79d612",
        "source_sha256": "5d2134ec54272858601ad55effb52536b2b7948f494b265ccca1b5abe1972374",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-08-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/08",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "08",
      "seed": 100001,
      "source_episode_id": "08-put-object-cabinet-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 141.26800670986995,
      "wall_time_s": 1917.646005173,
      "timing": {
        "wall_time_s": 1917.646005173,
        "simulation_time_s": 141.26800670986995,
        "physics_steps": 35317,
        "wall_time_start_ns": 1794764178665996418,
        "wall_time_end_ns": 1794766096312004314,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
      "native_instruction": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-08-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "6e48b119b1a20a61564e3dd2033c351aacabd365c9861a33f3eba60adf08a60d",
        "source_sha256": "caca8ce4bb51459231459dcaa4bc61aa4767bf9b3d8d8dfb4187dabec651db65",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-09-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/09",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "09",
      "seed": 100000,
      "source_episode_id": "09-blocks-ranking-size-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 348.5360165545717,
      "wall_time_s": 1784.365754478,
      "timing": {
        "wall_time_s": 1784.365754478,
        "simulation_time_s": 348.5360165545717,
        "physics_steps": 87134,
        "wall_time_start_ns": 1794764203595158291,
        "wall_time_end_ns": 1794765987960914899,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place large block, medium block, and small block in order at the table's center.",
      "native_instruction": "Place large block, medium block, and small block in order at the table's center.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-09-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "bc8061683ad606e323b7de48f620e290d95318bca486113c5fc5382b2992f861",
        "source_sha256": "6561d0b894e298541a7915b2b54c8ce02ee809fe106f94b3fd5d292481446633",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-10-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/10",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "10",
      "seed": 100000,
      "source_episode_id": "10-lift-pot-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 64.68400307232514,
      "wall_time_s": 1005.379032899,
      "timing": {
        "wall_time_s": 1005.379032899,
        "simulation_time_s": 64.68400307232514,
        "physics_steps": 16171,
        "wall_time_start_ns": 1794764181873141996,
        "wall_time_end_ns": 1794765187252177618,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the dark gray cooking kitchenpot using both arms",
      "native_instruction": "Lift the dark gray cooking kitchenpot using both arms",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-10-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "4459fc3321d2a75c54a02271b5f788728bd76f40dfea90aa53710856a361776a",
        "source_sha256": "b292672f0fe95738ca753159545ef720677e43b47455320ef2e5d3eeb9e9dc21",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-11-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/11",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "11",
      "seed": 100000,
      "source_episode_id": "11-adjust-bottle-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 95.00400451244786,
      "wall_time_s": 863.361224296,
      "timing": {
        "wall_time_s": 863.361224296,
        "simulation_time_s": 95.00400451244786,
        "physics_steps": 23751,
        "wall_time_start_ns": 1794764215056148963,
        "wall_time_end_ns": 1794765078417377265,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the brown bottle upright",
      "native_instruction": "Use the left arm to grab the brown bottle upright",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-11-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "109894304bc1be6e69a656a6ba7d763846dfaff6597e8c56a76a407cf1dd3ce4",
        "source_sha256": "d9ecb1bd8c8b3a679386cae1d7a12cee93a9410210e0a07392b6648bb4e08455",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-12-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/12",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "12",
      "seed": 100000,
      "source_episode_id": "12-beat-block-hammer-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 82.79200393240899,
      "wall_time_s": 877.250279781,
      "timing": {
        "wall_time_s": 877.250279781,
        "simulation_time_s": 82.79200393240899,
        "physics_steps": 20698,
        "wall_time_start_ns": 1794764221575043329,
        "wall_time_end_ns": 1794765098825326576,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the hammer with black handle then beat block",
      "native_instruction": "Use the left arm to grab the hammer with black handle then beat block",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-12-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "b8222a814f42c4cd5d914a57446f3e1df71e2ab119819c77ec38e3dca1691640",
        "source_sha256": "e6f8c5a1f454c77352924da1c581468817fe5225c1208c526df4675a389c36f3",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-13-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/13",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "13",
      "seed": 100000,
      "source_episode_id": "13-blocks-ranking-rgb-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 438
      },
      "steps": 438,
      "simulation_time_s": 129.1600061347708,
      "wall_time_s": 1126.514060884,
      "timing": {
        "wall_time_s": 1126.514060884,
        "simulation_time_s": 129.1600061347708,
        "physics_steps": 32290,
        "wall_time_start_ns": 1794764221780193071,
        "wall_time_end_ns": 1794765348294259268,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
      "native_instruction": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 438 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-13-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 14,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.975,
        "native_frames": 439,
        "published_frames": 439,
        "decode_ok": true,
        "sha256": "e441b85056df8db44b9f74c16b7e3c0e1bc2e23fdcc86ef11a60466c8aed5c9b",
        "source_sha256": "26aed87ee810f2c3077f8791d003e6c2bbd554ffa5279b32cd2f1aa089704ecd",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-14-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/14",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "14",
      "seed": 100000,
      "source_episode_id": "14-click-alarmclock-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 63
      },
      "steps": 63,
      "simulation_time_s": 23.81200113100931,
      "wall_time_s": 104.361427707,
      "timing": {
        "wall_time_s": 104.361427707,
        "simulation_time_s": 23.81200113100931,
        "physics_steps": 5953,
        "wall_time_start_ns": 1794764221288768079,
        "wall_time_end_ns": 1794764325650210134,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Tap the center top button of the alarm-clock with beige screen",
      "native_instruction": "Tap the center top button of the alarm-clock with beige screen",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 63 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-14-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 2,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.6,
        "native_frames": 64,
        "published_frames": 64,
        "decode_ok": true,
        "sha256": "4b29b49857dca027a22df6de54bb54fa97b855e2d5adc1ec438fc992048b536d",
        "source_sha256": "7a55809136ea91347a7d49a1fb67ec4d8e3cc208f5115cb3dfd2f2441d60c099",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-15-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/15",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "15",
      "seed": 100000,
      "source_episode_id": "15-click-bell-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 58
      },
      "steps": 58,
      "simulation_time_s": 23.3840011106804,
      "wall_time_s": 127.37127329,
      "timing": {
        "wall_time_s": 127.37127329,
        "simulation_time_s": 23.3840011106804,
        "physics_steps": 5846,
        "wall_time_start_ns": 1794764243626887202,
        "wall_time_end_ns": 1794764370998170060,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to press the medium-sized bell's top center.",
      "native_instruction": "Use the left arm to press the medium-sized bell's top center.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 58 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-15-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 2,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.475,
        "native_frames": 59,
        "published_frames": 59,
        "decode_ok": true,
        "sha256": "6682abb3f2c5f61820e2428111b71a1fd8bad3032ca8e5ce08a4f51f0c1496e1",
        "source_sha256": "8e4c1db05fba0828cc31b8f7c37a88c26f2925cab89b85c99bc12576a157082f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-16-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/16",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "16",
      "seed": 100006,
      "source_episode_id": "16-dump-bin-bigbin-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 600
      },
      "steps": 600,
      "simulation_time_s": 223.91201063524932,
      "wall_time_s": 1818.651265391,
      "timing": {
        "wall_time_s": 1818.651265391,
        "simulation_time_s": 223.91201063524932,
        "physics_steps": 55978,
        "wall_time_start_ns": 1794764254723811270,
        "wall_time_end_ns": 1794766073375079888,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
      "native_instruction": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 600 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-16-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100006,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 15.0,
        "native_frames": 600,
        "published_frames": 600,
        "decode_ok": true,
        "sha256": "a8e9b0d317df127d1d51793183d4d6e8e70dd85e7e9ede4adf74cffd2dc25b60",
        "source_sha256": "aed64e19d81c9f8e408b0515f320b28c030cef8eda7fe29b9f180de321e3c96d",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-17-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/17",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "17",
      "seed": 100000,
      "source_episode_id": "17-grab-roller-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 76
      },
      "steps": 76,
      "simulation_time_s": 26.648001265712082,
      "wall_time_s": 175.543360194,
      "timing": {
        "wall_time_s": 175.543360194,
        "simulation_time_s": 26.648001265712082,
        "physics_steps": 6662,
        "wall_time_start_ns": 1794764247844056246,
        "wall_time_end_ns": 1794764423387424106,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use arms to grab the smooth wooden roller.",
      "native_instruction": "Use arms to grab the smooth wooden roller.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 76 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-17-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.925,
        "native_frames": 77,
        "published_frames": 77,
        "decode_ok": true,
        "sha256": "dc024ed14e2db3e43906679ea7dd6f4415ec4901f6870a2cfb4c309ac6c6c155",
        "source_sha256": "7448221ce0e28af7f293243b4dd38e31d0564062f9d00c84e99d0d62057c5666",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-18-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/18",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "18",
      "seed": 100000,
      "source_episode_id": "18-handover-mic-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 177
      },
      "steps": 177,
      "simulation_time_s": 53.52800254244357,
      "wall_time_s": 377.676172642,
      "timing": {
        "wall_time_s": 377.676172642,
        "simulation_time_s": 53.52800254244357,
        "physics_steps": 13382,
        "wall_time_start_ns": 1794764252756926147,
        "wall_time_end_ns": 1794764630433111136,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use one arm to hold the microphone used for audio input and hand it over.",
      "native_instruction": "Use one arm to hold the microphone used for audio input and hand it over.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 177 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-18-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 6,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 4.45,
        "native_frames": 178,
        "published_frames": 178,
        "decode_ok": true,
        "sha256": "ff506de9cef6a71748f4ec30ef1042c4668056c05909445ae9372ed27b757a93",
        "source_sha256": "9ba54f008dcd8dc85c50c1be91c257fad72a50fe26e74ff6859e7a1122dd488e",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-19-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/19",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "19",
      "seed": 100000,
      "source_episode_id": "19-move-can-pot-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 103.48400491522625,
      "wall_time_s": 1157.177461029,
      "timing": {
        "wall_time_s": 1157.177461029,
        "simulation_time_s": 103.48400491522625,
        "physics_steps": 25871,
        "wall_time_start_ns": 1794764295809715494,
        "wall_time_end_ns": 1794765452987179824,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
      "native_instruction": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-19-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "42236601b467f45cf2f9e8fffd5fd69f4af50cff9f4d12c7bede4677602c11d2",
        "source_sha256": "a99a3eff9fdfc30011c8606431d78e66639ed5183fc6c7c260a5d35cd91b1952",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-20-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/20",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "20",
      "seed": 100001,
      "source_episode_id": "20-move-pillbottle-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 109.50400520116091,
      "wall_time_s": 1121.327676221,
      "timing": {
        "wall_time_s": 1121.327676221,
        "simulation_time_s": 109.50400520116091,
        "physics_steps": 27376,
        "wall_time_start_ns": 1794764266832320387,
        "wall_time_end_ns": 1794765388160005596,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the white cylindrical bottle and set it on the pad",
      "native_instruction": "Grab the white cylindrical bottle and set it on the pad",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-20-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "5d7e29a0ab8d12b8e222ed7dc958af8995a7e801396c0a2c2f2b66cb41b0f867",
        "source_sha256": "abc61e82121d7f597fd3f69bb767490a8a7b84c9f2f4bb73d4b46eee9fbe6f03",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-21-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/21",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "21",
      "seed": 100000,
      "source_episode_id": "21-move-playingcard-away-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 191
      },
      "steps": 191,
      "simulation_time_s": 54.49200258823112,
      "wall_time_s": 410.074574817,
      "timing": {
        "wall_time_s": 410.074574817,
        "simulation_time_s": 54.49200258823112,
        "physics_steps": 13623,
        "wall_time_start_ns": 1794764296454028268,
        "wall_time_end_ns": 1794764706528609544,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the right arm to grab the hard black box for card deck and relocate it.",
      "native_instruction": "Use the right arm to grab the hard black box for card deck and relocate it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 191 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-21-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 6,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 4.8,
        "native_frames": 192,
        "published_frames": 192,
        "decode_ok": true,
        "sha256": "f3d8165f71071e4715252c68e9e86dc825962931d8f3ca7666ca11f13997d019",
        "source_sha256": "ca14af5f44e6fba6b123640ee9a19e34423df2e37d44ce9fac9cd9ec7a50872f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-22-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/22",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "22",
      "seed": 100000,
      "source_episode_id": "22-move-stapler-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 108.37200514739379,
      "wall_time_s": 1002.444749262,
      "timing": {
        "wall_time_s": 1002.444749262,
        "simulation_time_s": 108.37200514739379,
        "physics_steps": 27093,
        "wall_time_start_ns": 1794764297790907632,
        "wall_time_end_ns": 1794765300235660716,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to move the plastic stapler to Yellow mat",
      "native_instruction": "Use the left arm to move the plastic stapler to Yellow mat",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-22-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "f6ecb9bc87425016f8c3224d0869cdfcec233ef6ce7f03988e26483a79a5d128",
        "source_sha256": "2cbdb350276821084e1e3f718faec5a6d360641714b7d097e13b132738520061",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-23-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/23",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "23",
      "seed": 100000,
      "source_episode_id": "23-open-laptop-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 128.15200608689338,
      "wall_time_s": 1866.374381608,
      "timing": {
        "wall_time_s": 1866.374381608,
        "simulation_time_s": 128.15200608689338,
        "physics_steps": 32038,
        "wall_time_start_ns": 1794764310951532859,
        "wall_time_end_ns": 1794766177325919618,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Open the flat orange laptop with one motion.",
      "native_instruction": "Open the flat orange laptop with one motion.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-23-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "059390a29cbf3e2f2d07cbc9f134e2de6560ea9429aa8b2d40c1cdf47f31b017",
        "source_sha256": "e95d87339c853c4691f3f5a2b7cc0a0ce3f3c43bb44462018d7c3e215b0a9bd4",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-24-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/24",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "24",
      "seed": 100000,
      "source_episode_id": "24-open-microwave-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1500
      },
      "steps": 1500,
      "simulation_time_s": 363.96401728736237,
      "wall_time_s": 2284.789132231,
      "timing": {
        "wall_time_s": 2284.789132231,
        "simulation_time_s": 363.96401728736237,
        "physics_steps": 90991,
        "wall_time_start_ns": 1794764292592182315,
        "wall_time_end_ns": 1794766577381317877,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
      "native_instruction": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-24-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1500,
        "max_control_steps": 1500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 47,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 37.5,
        "native_frames": 1500,
        "published_frames": 1500,
        "decode_ok": true,
        "sha256": "4d586aa53f65835d6a3d99eec2e3f5594cad5fbac3bf41a428e2da94e25d5005",
        "source_sha256": "5623b63088781c174bef3c1630573c8b2303f86f293d896ddbeafecbfeae012f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184,
          1216,
          1248,
          1280,
          1312,
          1344,
          1376,
          1408,
          1440,
          1472
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-25-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/25",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "25",
      "seed": 100005,
      "source_episode_id": "25-pick-diverse-bottles-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 80.3960038186051,
      "wall_time_s": 1176.503561818,
      "timing": {
        "wall_time_s": 1176.503561818,
        "simulation_time_s": 80.3960038186051,
        "physics_steps": 20099,
        "wall_time_start_ns": 1794764352198769888,
        "wall_time_end_ns": 1794765528702335789,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
      "native_instruction": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-25-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100005,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "53f5e9abdef9fe617eee54731e54b1eec69f19eb4ae97fc222f63ab7faafc9c2",
        "source_sha256": "73a95600e4b2c5bd66b5bb1e93a0cc8d04f67e5bae9035a073ae677777793fd0",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-26-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/26",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "26",
      "seed": 100000,
      "source_episode_id": "26-pick-dual-bottles-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 81.75200388301164,
      "wall_time_s": 926.668441293,
      "timing": {
        "wall_time_s": 926.668441293,
        "simulation_time_s": 81.75200388301164,
        "physics_steps": 20438,
        "wall_time_start_ns": 1794764354175237092,
        "wall_time_end_ns": 1794765280843681931,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
      "native_instruction": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-26-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "de3361e6a3861a28ae450bdee68bd94fdd31c1d06fa85ddf0663e0a0b73489bf",
        "source_sha256": "adcd73861d1f7376ff4cac1611e226690bcdc0cc25acc5a86c192254afd6f8e3",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-27-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/27",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "27",
      "seed": 100000,
      "source_episode_id": "27-place-a2b-left-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 91.14800432929769,
      "wall_time_s": 988.19620852,
      "timing": {
        "wall_time_s": 988.19620852,
        "simulation_time_s": 91.14800432929769,
        "physics_steps": 22787,
        "wall_time_start_ns": 1794764350568652989,
        "wall_time_end_ns": 1794765338764864674,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
      "native_instruction": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-27-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "3d1be58b3e92177e4b05d170147ddc3507746d8e1ab0f618fef03313b557d83f",
        "source_sha256": "d0737d9707001cfbbf228c47a41402a1de6e82149d0b809b52c6ba893db0c923",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-28-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/28",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "28",
      "seed": 100000,
      "source_episode_id": "28-place-a2b-right-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 97.13600461371243,
      "wall_time_s": 1428.506094326,
      "timing": {
        "wall_time_s": 1428.506094326,
        "simulation_time_s": 97.13600461371243,
        "physics_steps": 24284,
        "wall_time_start_ns": 1794764371353984277,
        "wall_time_end_ns": 1794765799860081702,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
      "native_instruction": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-28-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "38b64a3c99915681c5e3684b0398211a383ac9664f5c691fbb9afa8cab3f196b",
        "source_sha256": "1b82a84cf45da9fa11ae8d23dca0bc931c8c34be9c4a4d3630797ccc8599a491",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-29-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/29",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "29",
      "seed": 100000,
      "source_episode_id": "29-place-bread-basket-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 214.47601018706337,
      "wall_time_s": 1191.107732286,
      "timing": {
        "wall_time_s": 1191.107732286,
        "simulation_time_s": 214.47601018706337,
        "physics_steps": 53619,
        "wall_time_start_ns": 1794764337464175464,
        "wall_time_end_ns": 1794765528571915126,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
      "native_instruction": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-29-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "1079d3bcfe42d1f124a027a54581bd3185884680a0ff2ebaeb44dac2cb5b5616",
        "source_sha256": "bb832ccc79f70e7166038a13ace3e94cbd2c0e28d1a43ff02081eab947fbdfdf",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-30-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/30",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "30",
      "seed": 100001,
      "source_episode_id": "30-place-bread-skillet-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 248
      },
      "steps": 248,
      "simulation_time_s": 62.19600295415148,
      "wall_time_s": 846.120958935,
      "timing": {
        "wall_time_s": 846.120958935,
        "simulation_time_s": 62.19600295415148,
        "physics_steps": 15549,
        "wall_time_start_ns": 1794764387087408744,
        "wall_time_end_ns": 1794765233208373758,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
      "native_instruction": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 248 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-30-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 8,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 6.225,
        "native_frames": 249,
        "published_frames": 249,
        "decode_ok": true,
        "sha256": "8760ec68b0d0bc8a18e7f052cc6f5c9cbe547d286999cb35aa3e7a0cd68bd6a5",
        "source_sha256": "98c30e51db7fd8fa79feba55009c1dc923f16dd3abfce8d0ca25c37be51359a5",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-31-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/31",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "31",
      "seed": 100000,
      "source_episode_id": "31-place-burger-fries-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 138.9760066010058,
      "wall_time_s": 1329.373582868,
      "timing": {
        "wall_time_s": 1329.373582868,
        "simulation_time_s": 138.9760066010058,
        "physics_steps": 34744,
        "wall_time_start_ns": 1794764405717531649,
        "wall_time_end_ns": 1794765735091117792,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
      "native_instruction": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-31-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "c107416a6a76fd344234ca32807e0cb1d66a73d2f2c1fe8a683696cd044abc10",
        "source_sha256": "92a08217a2ab6dc2c189445c13692e309c3e3eeb48a84bbf7e48c9ea3ce4b309",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-32-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/32",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "32",
      "seed": 100000,
      "source_episode_id": "32-place-can-basket-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 466
      },
      "steps": 466,
      "simulation_time_s": 192.4080091388896,
      "wall_time_s": 1243.627200904,
      "timing": {
        "wall_time_s": 1243.627200904,
        "simulation_time_s": 192.4080091388896,
        "physics_steps": 48102,
        "wall_time_start_ns": 1794764391951826139,
        "wall_time_end_ns": 1794765635579031111,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "native_instruction": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 466 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-32-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 15,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 11.675,
        "native_frames": 467,
        "published_frames": 467,
        "decode_ok": true,
        "sha256": "b8f3aad812b748969eef4fc7eb2cb34d961e7b943b6aecd6da1398a0238642e8",
        "source_sha256": "dc6353b505afd51ba0947b79e850fedae3e9a4cba1ddd07da2a1ac1f1f8498f2",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-33-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/33",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "33",
      "seed": 100000,
      "source_episode_id": "33-place-cans-plasticbox-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 800
      },
      "steps": 800,
      "simulation_time_s": 241.27601145999506,
      "wall_time_s": 1128.715610553,
      "timing": {
        "wall_time_s": 1128.715610553,
        "simulation_time_s": 241.27601145999506,
        "physics_steps": 60319,
        "wall_time_start_ns": 1794764412830855414,
        "wall_time_end_ns": 1794765541546469327,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
      "native_instruction": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 800 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-33-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 25,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 20.0,
        "native_frames": 800,
        "published_frames": 800,
        "decode_ok": true,
        "sha256": "d03a748d2d51e6b3b1eeb16950151055890839de08cbb059e955975914bb2028",
        "source_sha256": "cfbd9516d25415f6deb11e53c07f76e635b9aad961fa0532ac11d6c2cf94ca8c",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-34-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/34",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "34",
      "seed": 100000,
      "source_episode_id": "34-place-container-plate-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 131
      },
      "steps": 131,
      "simulation_time_s": 49.28400234086439,
      "wall_time_s": 325.119621862,
      "timing": {
        "wall_time_s": 325.119621862,
        "simulation_time_s": 49.28400234086439,
        "physics_steps": 12321,
        "wall_time_start_ns": 1794764356508001350,
        "wall_time_end_ns": 1794764681627629708,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the curved bowl with smooth texture on the gray plate for dining.",
      "native_instruction": "Place the curved bowl with smooth texture on the gray plate for dining.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 131 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-34-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 5,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 3.3,
        "native_frames": 132,
        "published_frames": 132,
        "decode_ok": true,
        "sha256": "62285a1b5bad76ead021c2bbaf6abeb6854663cf320d60c4262d8f1b3691fb18",
        "source_sha256": "379e89f6a2c67e13f8bfa331bcb3c108cf45d0578a1933c626b6bfd462aaeb20",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-35-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/35",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "35",
      "seed": 100000,
      "source_episode_id": "35-place-empty-cup-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 127
      },
      "steps": 127,
      "simulation_time_s": 45.96400218317285,
      "wall_time_s": 411.557729656,
      "timing": {
        "wall_time_s": 411.557729656,
        "simulation_time_s": 45.96400218317285,
        "physics_steps": 11491,
        "wall_time_start_ns": 1794764465656848934,
        "wall_time_end_ns": 1794764877214584027,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
      "native_instruction": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 127 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-35-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 3.2,
        "native_frames": 128,
        "published_frames": 128,
        "decode_ok": true,
        "sha256": "9d3683d24936755b6d8170d73d3949c7019242e4982393fcd783e3f7ba74383a",
        "source_sha256": "2b95e9db26321e213858ac9fd6c7cbee21de5ff9f765c99131afbb7c174e40f5",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-36-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/36",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "36",
      "seed": 100000,
      "source_episode_id": "36-place-fan-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 116.48000553250313,
      "wall_time_s": 1155.021135959,
      "timing": {
        "wall_time_s": 1155.021135959,
        "simulation_time_s": 116.48000553250313,
        "physics_steps": 29120,
        "wall_time_start_ns": 1794764493119884124,
        "wall_time_end_ns": 1794765648141024333,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
      "native_instruction": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-36-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "14b7a5ef309f0eda41ad891dca34ccc8f6f00f660362b44a97f5cd926d17b2a9",
        "source_sha256": "21938ce60a695fe6c98fca83ca1343a2f4cb6b7bad8695c3b6c405b9039d8c0b",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-37-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/37",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "37",
      "seed": 100000,
      "source_episode_id": "37-place-mouse-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 106.4440050558187,
      "wall_time_s": 902.630015294,
      "timing": {
        "wall_time_s": 902.630015294,
        "simulation_time_s": 106.4440050558187,
        "physics_steps": 26611,
        "wall_time_start_ns": 1794764373800213620,
        "wall_time_end_ns": 1794765276430232928,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
      "native_instruction": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-37-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "92d4636a4a264d0bdfe6a889786c99b8e409d15c3eac5067d8ecddfd8d4e8c8d",
        "source_sha256": "11f38b534a8edf9de83c4445fe0ddb6bc772bdb41ad4181fb8c50db73abbbdf5",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-38-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/38",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "38",
      "seed": 100000,
      "source_episode_id": "38-place-object-basket-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 357
      },
      "steps": 357,
      "simulation_time_s": 127.96400607796386,
      "wall_time_s": 1342.627794537,
      "timing": {
        "wall_time_s": 1342.627794537,
        "simulation_time_s": 127.96400607796386,
        "physics_steps": 31991,
        "wall_time_start_ns": 1794764510374812556,
        "wall_time_end_ns": 1794765853002612844,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
      "native_instruction": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 357 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-38-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 12,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 8.95,
        "native_frames": 358,
        "published_frames": 358,
        "decode_ok": true,
        "sha256": "8333379c47704962c0570fe4142fe7da4dfa2b6b22d8c7044154103a12a87009",
        "source_sha256": "9583d23cdcfa37ae8ed12b57df1be16d294285fdcc2b938382801a5de2d5094d",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-39-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/39",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "39",
      "seed": 100000,
      "source_episode_id": "39-place-object-scale-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 102.74000487988815,
      "wall_time_s": 1063.210049673,
      "timing": {
        "wall_time_s": 1063.210049673,
        "simulation_time_s": 102.74000487988815,
        "physics_steps": 25685,
        "wall_time_start_ns": 1794764517984176737,
        "wall_time_end_ns": 1794765581194230423,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
      "native_instruction": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-39-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "b0c5dd66bbfcf646723309617ddcce40be7ff706235dcf06180d11599f49f823",
        "source_sha256": "7c8256b7e0995c96e5cf1ae6a93217e0ae33a7c480c21c7ddc7701827b71c070",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-40-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/40",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "40",
      "seed": 100001,
      "source_episode_id": "40-place-object-stand-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 117
      },
      "steps": 117,
      "simulation_time_s": 36.428001730237156,
      "wall_time_s": 340.449364091,
      "timing": {
        "wall_time_s": 340.449364091,
        "simulation_time_s": 36.428001730237156,
        "physics_steps": 9107,
        "wall_time_start_ns": 1794764545827850290,
        "wall_time_end_ns": 1794764886277219665,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the white stapler on the compact black hollow displaystand with the left arm.",
      "native_instruction": "Place the white stapler on the compact black hollow displaystand with the left arm.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 117 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-40-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 2.95,
        "native_frames": 118,
        "published_frames": 118,
        "decode_ok": true,
        "sha256": "8368eff65cb919e7b7fcf7fdd2a9010ebc3c8563c1480b0846f3bd7c972e9920",
        "source_sha256": "2696ecda4d8ff4410301f22b6937785e456f0595c10722a34830590f8f1a3693",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-41-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/41",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "41",
      "seed": 100000,
      "source_episode_id": "41-place-phone-stand-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 123.87200588360429,
      "wall_time_s": 820.339123202,
      "timing": {
        "wall_time_s": 820.339123202,
        "simulation_time_s": 123.87200588360429,
        "physics_steps": 30968,
        "wall_time_start_ns": 1794764575535113866,
        "wall_time_end_ns": 1794765395874241744,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Set the phone with flat black screen on the angled brown phonestand.",
      "native_instruction": "Set the phone with flat black screen on the angled brown phonestand.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-41-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "952667dfea8d7cc24100c91c5ef8b60b6202e4fbaa4d2e0fa57dd2dfa5dc3ae1",
        "source_sha256": "03bb1ff903004b1f42feae87eb1abb358c73a9217edd1128f4f887af72a5a993",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-42-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/42",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "42",
      "seed": 100001,
      "source_episode_id": "42-place-shoe-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 140
      },
      "steps": 140,
      "simulation_time_s": 44.92000213358551,
      "wall_time_s": 477.225878587,
      "timing": {
        "wall_time_s": 477.225878587,
        "simulation_time_s": 44.92000213358551,
        "physics_steps": 11230,
        "wall_time_start_ns": 1794764593858273466,
        "wall_time_end_ns": 1794765071084158326,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the medium adult shoe from the table and set it on the mat",
      "native_instruction": "Pick up the medium adult shoe from the table and set it on the mat",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 140 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-42-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 5,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 3.525,
        "native_frames": 141,
        "published_frames": 141,
        "decode_ok": true,
        "sha256": "cb31c778678166146d5e2a4c74dc1a75d31db9948739df3b5fa4d2ecfa1d5178",
        "source_sha256": "4790fe9caa99795b920a3d703f9033f1b9007bcab219a876476d7a2dc51f4ee4",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-43-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/43",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "43",
      "seed": 100000,
      "source_episode_id": "43-press-stapler-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 67.98800322925672,
      "wall_time_s": 975.684244367,
      "timing": {
        "wall_time_s": 975.684244367,
        "simulation_time_s": 67.98800322925672,
        "physics_steps": 16997,
        "wall_time_start_ns": 1794764700546541616,
        "wall_time_end_ns": 1794765676230789430,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Press the plastic stapler firmly using your arm.",
      "native_instruction": "Press the plastic stapler firmly using your arm.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-43-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "c1901032d1ba23eeb108a2d44f7b0fb9db4c30cc7cbe2a950e40168ef06fdb42",
        "source_sha256": "9f2edd47813791edece81d5dbb971f6f54ad4892140112e1a798c2a7357de52e",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-44-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/44",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "44",
      "seed": 100000,
      "source_episode_id": "44-rotate-qrcode-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 115.71600549621508,
      "wall_time_s": 867.641770998,
      "timing": {
        "wall_time_s": 867.641770998,
        "simulation_time_s": 115.71600549621508,
        "physics_steps": 28929,
        "wall_time_start_ns": 1794764610719792741,
        "wall_time_end_ns": 1794765478361568071,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
      "native_instruction": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-44-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "6c245948a8c07b4607156e682758c63c0667299e617c1ea0332605d3346d564f",
        "source_sha256": "8c813152afd42a53f90abd498eb91dd5f22893d35c83b14ecbe7f2603b92de26",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-45-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/45",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "45",
      "seed": 100001,
      "source_episode_id": "45-shake-bottle-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 81
      },
      "steps": 81,
      "simulation_time_s": 24.568001166917384,
      "wall_time_s": 335.825939497,
      "timing": {
        "wall_time_s": 335.825939497,
        "simulation_time_s": 24.568001166917384,
        "physics_steps": 6142,
        "wall_time_start_ns": 1794764608875915266,
        "wall_time_end_ns": 1794764944701861162,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the smooth bottle with blue text and shake it.",
      "native_instruction": "Pick up the smooth bottle with blue text and shake it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 81 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-45-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 2.05,
        "native_frames": 82,
        "published_frames": 82,
        "decode_ok": true,
        "sha256": "f215ed97ea10890a2bd55a39d2f34deaef23b9b95bfb4fb5fd147ad6099abc32",
        "source_sha256": "1c2dcfe452846f3b7bdc69e3b257d2ba3af628ccd5a0a8ce4947f548c84f7d68",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-46-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/46",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "46",
      "seed": 100001,
      "source_episode_id": "46-shake-bottle-horizontally-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 71
      },
      "steps": 71,
      "simulation_time_s": 24.680001172237098,
      "wall_time_s": 347.890419284,
      "timing": {
        "wall_time_s": 347.890419284,
        "simulation_time_s": 24.680001172237098,
        "physics_steps": 6170,
        "wall_time_start_ns": 1794764713590785203,
        "wall_time_end_ns": 1794765061481217473,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
      "native_instruction": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 71 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-46-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.8,
        "native_frames": 72,
        "published_frames": 72,
        "decode_ok": true,
        "sha256": "3bf0fed235b371f0243fff8b7908a39415549d87e8b1c64c421579747b0321e9",
        "source_sha256": "2a6b488a2be7a32733316b5b8a5209f4ec50d6588878c71d56805e3706d54f45",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-47-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/47",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "47",
      "seed": 100000,
      "source_episode_id": "47-stack-blocks-two-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 274
      },
      "steps": 274,
      "simulation_time_s": 84.32800400536507,
      "wall_time_s": 594.335999586,
      "timing": {
        "wall_time_s": 594.335999586,
        "simulation_time_s": 84.32800400536507,
        "physics_steps": 21082,
        "wall_time_start_ns": 1794764671325648003,
        "wall_time_end_ns": 1794765265661652757,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Move red block to the center, then put green block on top.",
      "native_instruction": "Move red block to the center, then put green block on top.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 274 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-47-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 9,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 6.875,
        "native_frames": 275,
        "published_frames": 275,
        "decode_ok": true,
        "sha256": "5dcd96923453f408ace6e4c8e317c07d24e30aa9dc9a58d9f1f71fd78113dd28",
        "source_sha256": "4d2eabebf88bbd4d5e4778f1701ac7aeec0597a838c673e40216bd19b532ef2c",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-48-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/48",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "48",
      "seed": 100000,
      "source_episode_id": "48-stack-bowls-two-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 316
      },
      "steps": 316,
      "simulation_time_s": 91.95200436748564,
      "wall_time_s": 1205.074599164,
      "timing": {
        "wall_time_s": 1205.074599164,
        "simulation_time_s": 91.95200436748564,
        "physics_steps": 22988,
        "wall_time_start_ns": 1794764721439816105,
        "wall_time_end_ns": 1794765926514420165,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
      "native_instruction": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 316 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-48-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 10,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 7.925,
        "native_frames": 317,
        "published_frames": 317,
        "decode_ok": true,
        "sha256": "65ebd1d6098fb88223e2ac71c0411e2d06eab8a3d96c12f50dbfb21b1cffd222",
        "source_sha256": "9e3635d7b10d76bfb518e9b272eea30abd179299dc013f170bdcd152b7c29cee",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-49-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/49",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "49",
      "seed": 100003,
      "source_episode_id": "49-stamp-seal-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 117.72000559139997,
      "wall_time_s": 1015.623128291,
      "timing": {
        "wall_time_s": 1015.623128291,
        "simulation_time_s": 117.72000559139997,
        "physics_steps": 29430,
        "wall_time_start_ns": 1794764727640435718,
        "wall_time_end_ns": 1794765743263566415,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the block-shaped seal with angled edges, press onto Blue",
      "native_instruction": "Grab the block-shaped seal with angled edges, press onto Blue",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-49-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100003,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "981d705bdaffb4fed9aac9a699528c341b7296c23642ce465b47cb38da65cddb",
        "source_sha256": "5dd5d5adb42b8f0dfbfc5a41c04ed014590c578b776b983c57e8974973f74e93",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-baseline-image_only-50-20261113",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/50",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "50",
      "seed": 100000,
      "source_episode_id": "50-turn-switch-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 84.22000400023535,
      "wall_time_s": 887.418108501,
      "timing": {
        "wall_time_s": 887.418108501,
        "simulation_time_s": 84.22000400023535,
        "physics_steps": 21055,
        "wall_time_start_ns": 1794764801403968109,
        "wall_time_end_ns": 1794765688822079784,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Press the flat square gray switch using the left arm",
      "native_instruction": "Press the flat square gray switch using the left arm",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/experiments/pose-standalone-native-hard-20261113/hard-native-baseline-image_only-50-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "98a239fc279dec653ca11986f52bf470cb42cd2374280df69945ffdc64577cc4",
        "source_sha256": "a2dbf8fae5fb75b9a80b7d64184ea108448c08397a409426564f96534e7207bf",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "native-full-kinex-pose-base-04-20261113",
      "variant": "kinex-pose-base",
      "task_key": "task03/04",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "04",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 11.6,
        "step_limit": 500,
        "steps": 290,
        "success": false,
        "valid": true
      },
      "steps": 290,
      "simulation_time_s": 11.6,
      "wall_time_s": 540.41621,
      "timing": {
        "simulation_time_s": 11.6,
        "wall_time_s": 540.41621,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "native_instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9528996651587558,
        "cache_read_input_tokens": 762112,
        "cache_reported_input_tokens": 799782,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 799782,
        "cached_input_tokens": 762112,
        "cost_usd": null,
        "fresh_input_tokens": 37670,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 799782,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 762112,
        "known_fresh_input_tokens": 37670,
        "known_input_tokens": 799782,
        "known_output_tokens": 6074,
        "known_reasoning_output_tokens": 1820,
        "output_tokens": 6074,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1820,
        "reasoning_reported_output_tokens": 6074,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 32,
          "cache_write_input_tokens": 32,
          "cache_write_reported_input_tokens": 32,
          "cached_input_tokens": 32,
          "fresh_input_tokens": 32,
          "input_tokens": 32,
          "output_tokens": 32,
          "reasoning_output_tokens": 32,
          "reasoning_reported_output_tokens": 32
        },
        "request_attempts": 32,
        "response_count": 32,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 805856,
        "uncached_input_tokens": 37670,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11830,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11870,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12003,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12051,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12344,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18043,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18613,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 18843,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 20345,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 20536,
            "output_tokens": 86,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20832,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21240,
            "output_tokens": 322,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 22702,
            "output_tokens": 111,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23279,
            "output_tokens": 159,
            "reasoning_output_tokens": 117
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 24617,
            "output_tokens": 849,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25488,
            "output_tokens": 130,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25724,
            "output_tokens": 139,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26376,
            "output_tokens": 52,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26490,
            "output_tokens": 103,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26621,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27458,
            "output_tokens": 232,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28498,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 29366,
            "output_tokens": 149,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30310,
            "output_tokens": 123,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31973,
            "output_tokens": 444,
            "reasoning_output_tokens": 354
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32485,
            "output_tokens": 106,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33400,
            "output_tokens": 560,
            "reasoning_output_tokens": 267
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 35171,
            "output_tokens": 445,
            "reasoning_output_tokens": 331
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36302,
            "output_tokens": 278,
            "reasoning_output_tokens": 122
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37378,
            "output_tokens": 178,
            "reasoning_output_tokens": 113
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 38365,
            "output_tokens": 842,
            "reasoning_output_tokens": 209
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39229,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 39,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "bash": 1,
          "devices": 1,
          "exec": 14,
          "device": 1,
          "write": 2,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 290 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; full 50-task suite."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/transcript.json",
        "usage": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/usage.json",
        "analysis": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/analysis.json",
        "protocol": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/instruction.json",
        "verdict": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/episode.json",
        "media_check": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/evidence.zip",
        "timing": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/timing.json",
        "native_session": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/session.jsonl",
        "trace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/scan-object.md",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/workspace/.kinex/memos/scan-object.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_cartesian.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-04-20261113/workspace/.kinex/tools/dual_cartesian.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 74,
        "observed_images": 9
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-progress-off-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 117,
        "published_frames": 117,
        "decode_ok": true,
        "duration_s": 2.925,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "dfa0ae67f972c8438bd01cdefddfc33f8c988319d471c02e206031f3b064a4eb",
        "source_sha256": {
          "third_person": "1531ad73fb125b537160c3730a95d6b161303b72b6e6474d33d5ffa48504fbf4",
          "left_wrist": "5e48b3afa705038077f3184dd3bc26f31ac49cb94f2e5ce0ef5229e43345863a",
          "right_wrist": "93bbdf9c4d96792a09360df60162fed6750ba8db0501962b5346c0a44af0bbd8"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "b56aab7a3603",
          "task": "robotwin/scan-object-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 118,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 117,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 118,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 117,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 118,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 117,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-progress-off-20261113",
      "vla_model": "progVLA-yz",
      "vla_used": false
    },
    {
      "id": "native-full-kinex-pose-base-01-20261113",
      "variant": "kinex-pose-base",
      "task_key": "task03/01",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "01",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 23.72,
        "step_limit": 800,
        "steps": 593,
        "success": true,
        "valid": true
      },
      "steps": 593,
      "simulation_time_s": 23.72,
      "wall_time_s": 1509.025507,
      "timing": {
        "simulation_time_s": 23.72,
        "wall_time_s": 1509.025507,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "native_instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9726336296459412,
        "cache_read_input_tokens": 2482304,
        "cache_reported_input_tokens": 2552147,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2552147,
        "cached_input_tokens": 2482304,
        "cost_usd": null,
        "fresh_input_tokens": 69843,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2552147,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2482304,
        "known_fresh_input_tokens": 69843,
        "known_input_tokens": 2552147,
        "known_output_tokens": 12484,
        "known_reasoning_output_tokens": 4508,
        "output_tokens": 12484,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 4508,
        "reasoning_reported_output_tokens": 12484,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 65,
          "cache_write_input_tokens": 65,
          "cache_write_reported_input_tokens": 65,
          "cached_input_tokens": 65,
          "fresh_input_tokens": 65,
          "input_tokens": 65,
          "output_tokens": 65,
          "reasoning_output_tokens": 65,
          "reasoning_reported_output_tokens": 65
        },
        "request_attempts": 65,
        "response_count": 65,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2564631,
        "uncached_input_tokens": 69843,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11825,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11865,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11998,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12048,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14486,
            "output_tokens": 238,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14592,
            "input_tokens": 19413,
            "output_tokens": 208,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 22483,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22777,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23442,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23634,
            "output_tokens": 162,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24837,
            "output_tokens": 428,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26032,
            "output_tokens": 479,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 27006,
            "output_tokens": 90,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 28275,
            "output_tokens": 311,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28778,
            "output_tokens": 1333,
            "reasoning_output_tokens": 355
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30134,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30280,
            "output_tokens": 46,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30388,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30640,
            "output_tokens": 64,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30875,
            "output_tokens": 202,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31106,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31833,
            "output_tokens": 119,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32590,
            "output_tokens": 97,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32709,
            "output_tokens": 143,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33486,
            "output_tokens": 159,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34287,
            "output_tokens": 129,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 35056,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35782,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36470,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 37196,
            "output_tokens": 398,
            "reasoning_output_tokens": 273
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 38287,
            "output_tokens": 110,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 39043,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39835,
            "output_tokens": 226,
            "reasoning_output_tokens": 186
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40756,
            "output_tokens": 213,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41342,
            "output_tokens": 125,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 42083,
            "output_tokens": 102,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42786,
            "output_tokens": 368,
            "reasoning_output_tokens": 320
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 43176,
            "output_tokens": 145,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 44015,
            "output_tokens": 69,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44782,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 44900,
            "output_tokens": 104,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45731,
            "output_tokens": 157,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46252,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46986,
            "output_tokens": 147,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47742,
            "output_tokens": 278,
            "reasoning_output_tokens": 239
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47872,
            "input_tokens": 48082,
            "output_tokens": 129,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48128,
            "input_tokens": 48826,
            "output_tokens": 611,
            "reasoning_output_tokens": 515
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49280,
            "input_tokens": 50164,
            "output_tokens": 118,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50176,
            "input_tokens": 51008,
            "output_tokens": 174,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51072,
            "input_tokens": 51548,
            "output_tokens": 306,
            "reasoning_output_tokens": 241
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51712,
            "input_tokens": 51994,
            "output_tokens": 235,
            "reasoning_output_tokens": 166
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52096,
            "input_tokens": 52258,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52224,
            "input_tokens": 53096,
            "output_tokens": 338,
            "reasoning_output_tokens": 255
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53248,
            "input_tokens": 54052,
            "output_tokens": 75,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54016,
            "input_tokens": 54742,
            "output_tokens": 96,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54656,
            "input_tokens": 55447,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55296,
            "input_tokens": 56156,
            "output_tokens": 680,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56704,
            "input_tokens": 57212,
            "output_tokens": 115,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57216,
            "input_tokens": 57951,
            "output_tokens": 320,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58240,
            "input_tokens": 58448,
            "output_tokens": 232,
            "reasoning_output_tokens": 145
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58624,
            "input_tokens": 59301,
            "output_tokens": 125,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59264,
            "input_tokens": 60045,
            "output_tokens": 79,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60032,
            "input_tokens": 60266,
            "output_tokens": 82,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60160,
            "input_tokens": 60608,
            "output_tokens": 863,
            "reasoning_output_tokens": 100
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61312,
            "input_tokens": 61496,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 78,
        "model_tool_calls_by_name": {
          "ls": 5,
          "read": 18,
          "find": 1,
          "devices": 3,
          "exec": 40,
          "device": 2,
          "write": 2,
          "device_stop": 4,
          "edit": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 593 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; full 50-task suite."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/transcript.json",
        "usage": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/usage.json",
        "analysis": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/analysis.json",
        "protocol": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/instruction.json",
        "verdict": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/episode.json",
        "media_check": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/evidence.zip",
        "timing": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/timing.json",
        "native_session": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/session.jsonl",
        "trace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/handover-block.md",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/workspace/.kinex/memos/handover-block.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-01-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 146,
        "observed_images": 24
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-progress-off-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 238,
        "published_frames": 238,
        "decode_ok": true,
        "duration_s": 5.95,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "5be60a83dae1a1f62404dfed24897dd10d62c4c02f50e2489f902391f8a129f2",
        "source_sha256": {
          "third_person": "a50ec31de073a56053358bdf0d172dc103b22297e8386f91e2c8aec2480d203e",
          "left_wrist": "d43821f0c4bbee1f9a26656e98cfb7ad31f9df7aa396a852ead75d1cc5503b46",
          "right_wrist": "37a9b2c8cb43e3f47dbcb7770c4e077eb54dcc7196b5e6f0c1217fc7f2c94ab3"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "858c590fb85f",
          "task": "robotwin/handover-block-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 239,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 238,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23720000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 239,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 238,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23720000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 239,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 238,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23720000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-progress-off-20261113",
      "vla_model": "progVLA-yz",
      "vla_used": false
    },
    {
      "id": "native-full-kinex-pose-base-02-20261113",
      "variant": "kinex-pose-base",
      "task_key": "task03/02",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "02",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 17.56,
        "step_limit": 600,
        "steps": 439,
        "success": false,
        "valid": true
      },
      "steps": 439,
      "simulation_time_s": 17.56,
      "wall_time_s": 1508.501615,
      "timing": {
        "simulation_time_s": 17.56,
        "wall_time_s": 1508.501615,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "native_instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9275536694872237,
        "cache_read_input_tokens": 1085440,
        "cache_reported_input_tokens": 1170218,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1170218,
        "cached_input_tokens": 1085440,
        "cost_usd": null,
        "fresh_input_tokens": 84778,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1170218,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1085440,
        "known_fresh_input_tokens": 84778,
        "known_input_tokens": 1170218,
        "known_output_tokens": 9198,
        "known_reasoning_output_tokens": 3296,
        "output_tokens": 9198,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3296,
        "reasoning_reported_output_tokens": 9198,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 38,
          "cache_write_input_tokens": 38,
          "cache_write_reported_input_tokens": 38,
          "cached_input_tokens": 38,
          "fresh_input_tokens": 38,
          "input_tokens": 38,
          "output_tokens": 38,
          "reasoning_output_tokens": 38,
          "reasoning_reported_output_tokens": 38
        },
        "request_attempts": 38,
        "response_count": 38,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1179416,
        "uncached_input_tokens": 84778,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11819,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11952,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12000,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12293,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 15027,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 15236,
            "output_tokens": 288,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14976,
            "input_tokens": 21499,
            "output_tokens": 96,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15360,
            "input_tokens": 21747,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 21937,
            "output_tokens": 163,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 24178,
            "output_tokens": 386,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25962,
            "output_tokens": 617,
            "reasoning_output_tokens": 212
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26978,
            "output_tokens": 286,
            "reasoning_output_tokens": 244
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 28443,
            "output_tokens": 1057,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29521,
            "output_tokens": 211,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 29885,
            "output_tokens": 236,
            "reasoning_output_tokens": 104
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30647,
            "output_tokens": 143,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31328,
            "output_tokens": 81,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 31892,
            "output_tokens": 183,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 32606,
            "output_tokens": 388,
            "reasoning_output_tokens": 306
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33500,
            "output_tokens": 146,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34153,
            "output_tokens": 65,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34240,
            "output_tokens": 459,
            "reasoning_output_tokens": 405
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35182,
            "output_tokens": 102,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35795,
            "output_tokens": 245,
            "reasoning_output_tokens": 130
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36100,
            "output_tokens": 58,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36220,
            "output_tokens": 151,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36831,
            "output_tokens": 287,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37437,
            "output_tokens": 669,
            "reasoning_output_tokens": 463
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38572,
            "output_tokens": 409,
            "reasoning_output_tokens": 130
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39179,
            "output_tokens": 337,
            "reasoning_output_tokens": 244
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 40029,
            "output_tokens": 348,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40575,
            "output_tokens": 302,
            "reasoning_output_tokens": 216
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41389,
            "output_tokens": 107,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41985,
            "output_tokens": 144,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42641,
            "output_tokens": 101,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43259,
            "output_tokens": 727,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 44010,
            "output_tokens": 134,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44171,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 48,
        "model_tool_calls_by_name": {
          "read": 15,
          "ls": 1,
          "bash": 2,
          "exec": 23,
          "devices": 1,
          "device": 1,
          "write": 2,
          "device_stop": 2,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 439 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; full 50-task suite."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/transcript.json",
        "usage": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/usage.json",
        "analysis": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/analysis.json",
        "protocol": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/instruction.json",
        "verdict": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/episode.json",
        "media_check": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/evidence.zip",
        "timing": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/timing.json",
        "native_session": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/session.jsonl",
        "trace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-shoes.md",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/workspace/.kinex/memos/robotwin-shoes.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-02-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 89,
        "observed_images": 15
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-progress-off-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 177,
        "published_frames": 177,
        "decode_ok": true,
        "duration_s": 4.425,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "66f6d8fdfd704baf68c46cc796930b4b13078851ec425bce43589d4a5c6dd6d1",
        "source_sha256": {
          "third_person": "97e05253bdeec32df37f35e851d570265d241edfef118eb5e11d43d45f543299",
          "left_wrist": "67f03d826d1d0ab6d06f601123177a653cb12b6f36d540fdffd38ecab402ba44",
          "right_wrist": "17ecab90064229ad46feda333d0514c13599fbe4e92a620a18bc20ac3d7b962c"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "d889bdb6e25d",
          "task": "robotwin/place-dual-shoes-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 177,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 177,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 177,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 177,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 177,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 177,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-progress-off-20261113",
      "vla_model": "progVLA-yz",
      "vla_used": false
    },
    {
      "id": "native-full-kinex-pose-base-06-20261113",
      "variant": "kinex-pose-base",
      "task_key": "task03/06",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "06",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 33.96,
        "step_limit": 1200,
        "steps": 849,
        "success": true,
        "valid": true
      },
      "steps": 849,
      "simulation_time_s": 33.96,
      "wall_time_s": 2011.901282,
      "timing": {
        "simulation_time_s": 33.96,
        "wall_time_s": 2011.901282,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "native_instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.968687681488997,
        "cache_read_input_tokens": 1329024,
        "cache_reported_input_tokens": 1371984,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1371984,
        "cached_input_tokens": 1329024,
        "cost_usd": null,
        "fresh_input_tokens": 42960,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1371984,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1329024,
        "known_fresh_input_tokens": 42960,
        "known_input_tokens": 1371984,
        "known_output_tokens": 8090,
        "known_reasoning_output_tokens": 1888,
        "output_tokens": 8090,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1888,
        "reasoning_reported_output_tokens": 8090,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 47,
          "cache_write_input_tokens": 47,
          "cache_write_reported_input_tokens": 47,
          "cached_input_tokens": 47,
          "fresh_input_tokens": 47,
          "input_tokens": 47,
          "output_tokens": 47,
          "reasoning_output_tokens": 47,
          "reasoning_reported_output_tokens": 47
        },
        "request_attempts": 47,
        "response_count": 47,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1380074,
        "uncached_input_tokens": 42960,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11821,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11861,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11994,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12044,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12337,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 12488,
            "output_tokens": 168,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 16409,
            "output_tokens": 147,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 19354,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 19584,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 20832,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21418,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21670,
            "output_tokens": 89,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 22128,
            "output_tokens": 299,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22807,
            "output_tokens": 445,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 24008,
            "output_tokens": 218,
            "reasoning_output_tokens": 176
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 25405,
            "output_tokens": 116,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25987,
            "output_tokens": 1273,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27281,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 28012,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28112,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28355,
            "output_tokens": 183,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28565,
            "output_tokens": 114,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29303,
            "output_tokens": 139,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30162,
            "output_tokens": 219,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30441,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30530,
            "output_tokens": 76,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30975,
            "output_tokens": 180,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31868,
            "output_tokens": 60,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32699,
            "output_tokens": 107,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 32833,
            "output_tokens": 101,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33654,
            "output_tokens": 194,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34659,
            "output_tokens": 255,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35662,
            "output_tokens": 49,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35914,
            "output_tokens": 246,
            "reasoning_output_tokens": 155
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36187,
            "output_tokens": 162,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 37103,
            "output_tokens": 220,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37350,
            "output_tokens": 162,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 38408,
            "output_tokens": 114,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 39293,
            "output_tokens": 250,
            "reasoning_output_tokens": 102
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 40288,
            "output_tokens": 172,
            "reasoning_output_tokens": 102
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 41433,
            "output_tokens": 87,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 42102,
            "output_tokens": 220,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42955,
            "output_tokens": 156,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43253,
            "output_tokens": 98,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43610,
            "output_tokens": 426,
            "reasoning_output_tokens": 255
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44063,
            "output_tokens": 681,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 44767,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 53,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 17,
          "bash": 1,
          "devices": 5,
          "exec": 19,
          "write": 2,
          "edit": 6,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 849 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; full 50-task suite."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/transcript.json",
        "usage": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/usage.json",
        "analysis": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/analysis.json",
        "protocol": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/instruction.json",
        "verdict": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/episode.json",
        "media_check": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/evidence.zip",
        "timing": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/timing.json",
        "native_session": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/session.jsonl",
        "trace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack-blocks.md",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/workspace/.kinex/memos/stack-blocks.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/stack_control.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-06-20261113/workspace/.kinex/tools/stack_control.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 103,
        "observed_images": 12
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-progress-off-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 341,
        "published_frames": 341,
        "decode_ok": true,
        "duration_s": 8.525,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "217182e92e0d3e305a1ec92fd74a20aae4aea5d6c9627d7f1b35a14032299809",
        "source_sha256": {
          "third_person": "957629d8ad264249e3e138485f84a9fa7fa343bebf23826c770d8567d01f36c1",
          "left_wrist": "ed0214c166bd512b5f6619a4f5a6672b6698d0f3e96188e1ab0e13d2d1fcd917",
          "right_wrist": "54ffbbae2e4039b31c4ae388e17b564a53a3f61c37acd5ec3f1387143dbeaec6"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "1afd842cad95",
          "task": "robotwin/stack-blocks-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 341,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 341,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 341,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 341,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 341,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 341,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-progress-off-20261113",
      "vla_model": "progVLA-yz",
      "vla_used": false
    },
    {
      "id": "native-full-kinex-pose-base-05-20261113",
      "variant": "kinex-pose-base",
      "task_key": "task03/05",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "05",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 37.480000000000004,
        "step_limit": 1200,
        "steps": 937,
        "success": true,
        "valid": true
      },
      "steps": 937,
      "simulation_time_s": 37.480000000000004,
      "wall_time_s": 2765.023235,
      "timing": {
        "simulation_time_s": 37.480000000000004,
        "wall_time_s": 2765.023235,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "native_instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9678197894708417,
        "cache_read_input_tokens": 2867200,
        "cache_reported_input_tokens": 2962535,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2962535,
        "cached_input_tokens": 2867200,
        "cost_usd": null,
        "fresh_input_tokens": 95335,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2962535,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2867200,
        "known_fresh_input_tokens": 95335,
        "known_input_tokens": 2962535,
        "known_output_tokens": 14232,
        "known_reasoning_output_tokens": 4097,
        "output_tokens": 14232,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 4097,
        "reasoning_reported_output_tokens": 14232,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 77,
          "cache_write_input_tokens": 77,
          "cache_write_reported_input_tokens": 77,
          "cached_input_tokens": 77,
          "fresh_input_tokens": 77,
          "input_tokens": 77,
          "output_tokens": 77,
          "reasoning_output_tokens": 77,
          "reasoning_reported_output_tokens": 77
        },
        "request_attempts": 77,
        "response_count": 77,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2976767,
        "uncached_input_tokens": 95335,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11821,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11861,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11994,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12042,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14188,
            "output_tokens": 224,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14208,
            "input_tokens": 19101,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 19671,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20778,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21096,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 22869,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23099,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23289,
            "output_tokens": 166,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 24496,
            "output_tokens": 281,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 26157,
            "output_tokens": 294,
            "reasoning_output_tokens": 255
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 27146,
            "output_tokens": 439,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28170,
            "output_tokens": 462,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28836,
            "output_tokens": 705,
            "reasoning_output_tokens": 197
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29562,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29933,
            "output_tokens": 1155,
            "reasoning_output_tokens": 299
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31109,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31487,
            "output_tokens": 124,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31911,
            "output_tokens": 256,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32480,
            "output_tokens": 181,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32842,
            "output_tokens": 146,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33293,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33673,
            "output_tokens": 211,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34189,
            "output_tokens": 298,
            "reasoning_output_tokens": 180
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34668,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35042,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35181,
            "output_tokens": 53,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35256,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35650,
            "output_tokens": 125,
            "reasoning_output_tokens": 100
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35944,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36796,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36853,
            "output_tokens": 104,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36984,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37434,
            "output_tokens": 187,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38341,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38399,
            "output_tokens": 65,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 39099,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39206,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39516,
            "output_tokens": 203,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39967,
            "output_tokens": 214,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40913,
            "output_tokens": 415,
            "reasoning_output_tokens": 112
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41476,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 43402,
            "output_tokens": 308,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 44076,
            "output_tokens": 488,
            "reasoning_output_tokens": 400
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44752,
            "output_tokens": 137,
            "reasoning_output_tokens": 112
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45059,
            "output_tokens": 224,
            "reasoning_output_tokens": 136
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 46016,
            "output_tokens": 511,
            "reasoning_output_tokens": 157
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 46896,
            "output_tokens": 244,
            "reasoning_output_tokens": 157
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 47332,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47360,
            "input_tokens": 47606,
            "output_tokens": 69,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 47844,
            "output_tokens": 196,
            "reasoning_output_tokens": 113
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48000,
            "input_tokens": 48232,
            "output_tokens": 58,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 48312,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 48506,
            "output_tokens": 254,
            "reasoning_output_tokens": 165
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48640,
            "input_tokens": 48949,
            "output_tokens": 140,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49277,
            "output_tokens": 156,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49280,
            "input_tokens": 49621,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 49813,
            "output_tokens": 140,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49792,
            "input_tokens": 50134,
            "output_tokens": 133,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 50456,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50432,
            "input_tokens": 50733,
            "output_tokens": 191,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50816,
            "input_tokens": 51123,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51072,
            "input_tokens": 51316,
            "output_tokens": 333,
            "reasoning_output_tokens": 227
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51456,
            "input_tokens": 51767,
            "output_tokens": 413,
            "reasoning_output_tokens": 192
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52096,
            "input_tokens": 52673,
            "output_tokens": 668,
            "reasoning_output_tokens": 292
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53248,
            "input_tokens": 54909,
            "output_tokens": 137,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54912,
            "input_tokens": 55235,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55168,
            "input_tokens": 55504,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55424,
            "input_tokens": 55698,
            "output_tokens": 182,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55680,
            "input_tokens": 56026,
            "output_tokens": 94,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55936,
            "input_tokens": 56348,
            "output_tokens": 301,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56448,
            "input_tokens": 56676,
            "output_tokens": 191,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56704,
            "input_tokens": 56894,
            "output_tokens": 614,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57344,
            "input_tokens": 57532,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 84,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 15,
          "find": 1,
          "exec": 43,
          "devices": 4,
          "device": 8,
          "write": 3,
          "device_stop": 4,
          "edit": 4
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 937 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; full 50-task suite."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/transcript.json",
        "usage": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/usage.json",
        "analysis": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/analysis.json",
        "protocol": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/instruction.json",
        "verdict": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/episode.json",
        "media_check": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/evidence.zip",
        "timing": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/timing.json",
        "native_session": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/session.jsonl",
        "trace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack_bowls.md",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/workspace/.kinex/memos/stack_bowls.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/bowl_geometry.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/workspace/.kinex/tools/bowl_geometry.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-pose-base-05-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 164,
        "observed_images": 18
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-progress-off-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 376,
        "published_frames": 376,
        "decode_ok": true,
        "duration_s": 9.4,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "ae88747525a0fbf78e4c4a2e89fc6cd7f4bda44df9bf807962e1dc0bcb675aa0",
        "source_sha256": {
          "third_person": "0eaa402975c88b9ca73bf3a54dda0231a5291e041380e6baa696207a39438f30",
          "left_wrist": "65decb7c0fbc481f652f09d99ed0e1f1da8dfd2839949e36c9ac16eb6cdee432",
          "right_wrist": "0f0908382074073da14c986ddd2de57430b2fb64a4340e0ed3b958dbbafe5237"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "a953e4c76f97",
          "task": "robotwin/stack-bowls-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 376,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 376,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 37480000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 376,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 376,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 37480000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 376,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 376,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 37480000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-progress-off-20261113",
      "vla_model": "progVLA-yz",
      "vla_used": false
    },
    {
      "id": "hard-native-jy-image_only-01-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/01",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "01",
      "seed": 100001,
      "source_episode_id": "01-handover-block-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 800
      },
      "steps": 800,
      "simulation_time_s": 185.3880088054575,
      "wall_time_s": 2055.29086187,
      "timing": {
        "wall_time_s": 2055.29086187,
        "simulation_time_s": 185.3880088054575,
        "physics_steps": 46347,
        "wall_time_start_ns": 1794687762598136988,
        "wall_time_end_ns": 1794689817889003456,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
      "native_instruction": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 800 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-01-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 25,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 20.0,
        "native_frames": 800,
        "published_frames": 800,
        "decode_ok": true,
        "sha256": "9c0d8fb74c5ed22bf7d57a366c58c4676dd137512aa01f5c682939bc7359e07a",
        "source_sha256": "ea42e8a09fb0d44cad9377f80d93b3e4076e5365b24be5c14b8b60fa82738936",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-02-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/02",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "02",
      "seed": 100001,
      "source_episode_id": "02-place-dual-shoes-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 600
      },
      "steps": 600,
      "simulation_time_s": 213.7160101509653,
      "wall_time_s": 1820.640017255,
      "timing": {
        "wall_time_s": 1820.640017255,
        "simulation_time_s": 213.7160101509653,
        "physics_steps": 53429,
        "wall_time_start_ns": 1794687782350657119,
        "wall_time_end_ns": 1794689602990677857,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
      "native_instruction": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 600 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-02-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 15.0,
        "native_frames": 600,
        "published_frames": 600,
        "decode_ok": true,
        "sha256": "20abd6238bbdf1dc7d9e1797b764c386101ec598c42364d6fefc590dead98715",
        "source_sha256": "bd148d95c85e416a4428b30a654dd17493f9190fa989aff2384f3cb72e685582",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-03-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/03",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "03",
      "seed": 100000,
      "source_episode_id": "03-put-bottles-dustbin-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1700
      },
      "steps": 1700,
      "simulation_time_s": 586.6600278648548,
      "wall_time_s": 4524.395932251,
      "timing": {
        "wall_time_s": 4524.395932251,
        "simulation_time_s": 586.6600278648548,
        "physics_steps": 146665,
        "wall_time_start_ns": 1794687803440179361,
        "wall_time_end_ns": 1794692327836114825,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
      "native_instruction": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-03-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1700,
        "max_control_steps": 1700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 54,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 42.5,
        "native_frames": 1700,
        "published_frames": 1700,
        "decode_ok": true,
        "sha256": "f31dc1537225c915bbf92d7e3fb1c3a2c3f13d20c97b780f88c77493468bf79a",
        "source_sha256": "03886759fc21b885ae5f45047072407728bf8f1977fb29eadd6e5ab444ec0b83",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184,
          1216,
          1248,
          1280,
          1312,
          1344,
          1376,
          1408,
          1440,
          1472,
          1504,
          1536,
          1568,
          1600,
          1632,
          1664,
          1696
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-04-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/04",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "04",
      "seed": 100001,
      "source_episode_id": "04-scan-object-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 160
      },
      "steps": 160,
      "simulation_time_s": 56.28000267315656,
      "wall_time_s": 358.972524202,
      "timing": {
        "wall_time_s": 358.972524202,
        "simulation_time_s": 56.28000267315656,
        "physics_steps": 14070,
        "wall_time_start_ns": 1794605910734400907,
        "wall_time_end_ns": 1794606269706930321,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
      "native_instruction": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 160 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-04-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 5,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 4.025,
        "native_frames": 161,
        "published_frames": 161,
        "decode_ok": true,
        "sha256": "390b20a912ca6c581780505442b5ee4fe5828e1d4fb3ca79683df611e3083af0",
        "source_sha256": "0157fc8ef75fd771a716347564d5787b4a04320f0b39bd2cdf28764fe5981135",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-05-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/05",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "05",
      "seed": 100000,
      "source_episode_id": "05-stack-bowls-three-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 335.9720159578137,
      "wall_time_s": 4096.166525205,
      "timing": {
        "wall_time_s": 4096.166525205,
        "simulation_time_s": 335.9720159578137,
        "physics_steps": 83993,
        "wall_time_start_ns": 1794687917928315553,
        "wall_time_end_ns": 1794692014094846216,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
      "native_instruction": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-05-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "19445d510e6f230f7f0ee55693c113fca7a6e2eb5525b106b54d6d4bfda0bc3c",
        "source_sha256": "0d77f120fd7be3e1a323e5b2243de1279d12b8552e6e4d41d884f2223281d9c8",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-06-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/06",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "06",
      "seed": 100000,
      "source_episode_id": "06-stack-blocks-three-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 313.58801489463076,
      "wall_time_s": 4085.377929141,
      "timing": {
        "wall_time_s": 4085.377929141,
        "simulation_time_s": 313.58801489463076,
        "physics_steps": 78397,
        "wall_time_start_ns": 1794688006337613985,
        "wall_time_end_ns": 1794692091715545862,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
      "native_instruction": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-06-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "d9d06b6089ddb90c8d223db24b03bfca375b53871556e1f225b24de5667a1bca",
        "source_sha256": "caca616bb6c7566aedca5d26bab3ca5141d6359b276ce3744ba81c8266a2a51f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-07-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/07",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "07",
      "seed": 100000,
      "source_episode_id": "07-hanging-mug-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 900
      },
      "steps": 900,
      "simulation_time_s": 202.60000962298363,
      "wall_time_s": 3152.694151735,
      "timing": {
        "wall_time_s": 3152.694151735,
        "simulation_time_s": 202.60000962298363,
        "physics_steps": 50650,
        "wall_time_start_ns": 1794687967724833858,
        "wall_time_end_ns": 1794691120418989746,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
      "native_instruction": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 900 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-07-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 29,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 22.5,
        "native_frames": 900,
        "published_frames": 900,
        "decode_ok": true,
        "sha256": "4f6e77afeb107dee5ebf2cef2cda657ff3a09dbb28118efba9b6aaba25986703",
        "source_sha256": "79a1edd5f65032313054529df23b6b1ce72a4876e8a513bfd3fc9cbf5ae68318",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-08-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/08",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "08",
      "seed": 100001,
      "source_episode_id": "08-put-object-cabinet-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 153.1720072752796,
      "wall_time_s": 2594.330067401,
      "timing": {
        "wall_time_s": 2594.330067401,
        "simulation_time_s": 153.1720072752796,
        "physics_steps": 38293,
        "wall_time_start_ns": 1794688070923748299,
        "wall_time_end_ns": 1794690665253819666,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
      "native_instruction": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-08-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "c1c5a5d3fcc295de7fcfffa82eb2294687d9c21a713a24f4539e94fa8008174b",
        "source_sha256": "0d912c9e11707d10f48bb2c684c3de887b1ac5aefcbe84000d2e15045e2d87f2",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-09-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/09",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "09",
      "seed": 100000,
      "source_episode_id": "09-blocks-ranking-size-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 302.896014386788,
      "wall_time_s": 3767.570066488,
      "timing": {
        "wall_time_s": 3767.570066488,
        "simulation_time_s": 302.896014386788,
        "physics_steps": 75724,
        "wall_time_start_ns": 1794688466391360104,
        "wall_time_end_ns": 1794692233961429118,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place large block, medium block, and small block in order at the table's center.",
      "native_instruction": "Place large block, medium block, and small block in order at the table's center.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-09-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "7984109c88aefc8bb9302f3df6799afc19670c392c7a073a1c80de8c43d568ad",
        "source_sha256": "bc3a19eb8f77dc14e816e4112004e8316fcf1109d49a6ae191488ccee40e64a6",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-10-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/10",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "10",
      "seed": 100000,
      "source_episode_id": "10-lift-pot-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 68.96400327561423,
      "wall_time_s": 1569.238550255,
      "timing": {
        "wall_time_s": 1569.238550255,
        "simulation_time_s": 68.96400327561423,
        "physics_steps": 17241,
        "wall_time_start_ns": 1794689945646377278,
        "wall_time_end_ns": 1794691514884932155,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the dark gray cooking kitchenpot using both arms",
      "native_instruction": "Lift the dark gray cooking kitchenpot using both arms",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-10-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "f7111acd00320a5f007fe63e767961293a0791b73575f7f5df8e369312546eab",
        "source_sha256": "d309c604a170e32eb64d7fc1d1be5d7367ee784ab86fa8082718017c83a1eeea",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-11-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/11",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "11",
      "seed": 100000,
      "source_episode_id": "11-adjust-bottle-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 99.33200471801683,
      "wall_time_s": 1263.22527186,
      "timing": {
        "wall_time_s": 1263.22527186,
        "simulation_time_s": 99.33200471801683,
        "physics_steps": 24833,
        "wall_time_start_ns": 1794690211076709763,
        "wall_time_end_ns": 1794691474301984454,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the brown bottle upright",
      "native_instruction": "Use the left arm to grab the brown bottle upright",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-11-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "52b3f4a00183282e14cdbad9e4d6debbd47dcc66d699cf8373eaa3950821b9f8",
        "source_sha256": "91ee8d8f2c48b35baaf0540b242623bb64194b8e756c3f20851fb35a5bfb153c",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-12-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/12",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "12",
      "seed": 100000,
      "source_episode_id": "12-beat-block-hammer-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 85.60400406597182,
      "wall_time_s": 465.995241671,
      "timing": {
        "wall_time_s": 465.995241671,
        "simulation_time_s": 85.60400406597182,
        "physics_steps": 21401,
        "wall_time_start_ns": 1794690277010932336,
        "wall_time_end_ns": 1794690743006177480,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the hammer with black handle then beat block",
      "native_instruction": "Use the left arm to grab the hammer with black handle then beat block",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-12-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "af8592fb4a7f31473288cc533cff04b97735be8458b3bf72d96bba7ec0293081",
        "source_sha256": "b8e8b983e5cce0e46da19ad5ade63ffc957d0b7db1950fee1ad5d661fee85f1b",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-13-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/13",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "13",
      "seed": 100000,
      "source_episode_id": "13-blocks-ranking-rgb-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 296.7040140926838,
      "wall_time_s": 1854.89296379,
      "timing": {
        "wall_time_s": 1854.89296379,
        "simulation_time_s": 296.7040140926838,
        "physics_steps": 74176,
        "wall_time_start_ns": 1794690291980559260,
        "wall_time_end_ns": 1794692146873525326,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
      "native_instruction": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-13-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "0f779054776ccc15f30fe553940e73919ef479d69cf5c87c4de905346037ddbe",
        "source_sha256": "b0d6fe8d5c1f5a66187eea0db4b7085662f95d1aab60c79bd0fb1cc94f9763e8",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-14-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/14",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "14",
      "seed": 100000,
      "source_episode_id": "14-click-alarmclock-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 64
      },
      "steps": 64,
      "simulation_time_s": 21.17200100561604,
      "wall_time_s": 157.477960373,
      "timing": {
        "wall_time_s": 157.477960373,
        "simulation_time_s": 21.17200100561604,
        "physics_steps": 5293,
        "wall_time_start_ns": 1794690365836166927,
        "wall_time_end_ns": 1794690523314133931,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Tap the center top button of the alarm-clock with beige screen",
      "native_instruction": "Tap the center top button of the alarm-clock with beige screen",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 64 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-14-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 2,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.625,
        "native_frames": 65,
        "published_frames": 65,
        "decode_ok": true,
        "sha256": "e6e989b283f99ed6f5aea4b37bae5df08a0a48d41aac696e46f9c43e03543a3a",
        "source_sha256": "fda1c22666d22450d0a470c450651e82816c5717f62d45cf4b6ca5329dda5f38",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-15-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/15",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "15",
      "seed": 100000,
      "source_episode_id": "15-click-bell-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 60
      },
      "steps": 60,
      "simulation_time_s": 23.20000110194087,
      "wall_time_s": 127.84284813,
      "timing": {
        "wall_time_s": 127.84284813,
        "simulation_time_s": 23.20000110194087,
        "physics_steps": 5800,
        "wall_time_start_ns": 1794690457318735041,
        "wall_time_end_ns": 1794690585161588413,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to press the medium-sized bell's top center.",
      "native_instruction": "Use the left arm to press the medium-sized bell's top center.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 60 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-15-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 2,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.525,
        "native_frames": 61,
        "published_frames": 61,
        "decode_ok": true,
        "sha256": "3bb2001fe44083e6c47bfc8e061b1dafbd56d134b9abafb7329b244fb577ba00",
        "source_sha256": "b934bc2774b51085bbb0864beec8762e6361abcec75958eb44e8103232fa6de3",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-16-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/16",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "16",
      "seed": 100006,
      "source_episode_id": "16-dump-bin-bigbin-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 600
      },
      "steps": 600,
      "simulation_time_s": 179.12400850793347,
      "wall_time_s": 1095.065109298,
      "timing": {
        "wall_time_s": 1095.065109298,
        "simulation_time_s": 179.12400850793347,
        "physics_steps": 44781,
        "wall_time_start_ns": 1794690847703811400,
        "wall_time_end_ns": 1794691942768924583,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
      "native_instruction": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 600 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-16-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100006,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 15.0,
        "native_frames": 600,
        "published_frames": 600,
        "decode_ok": true,
        "sha256": "246741e7ae02c2eec08b07bc484a539452f9c59748370ad5da4df6f3cfeb6561",
        "source_sha256": "efd74523ae2af1de84afe42b279b851a7874d55dfa25f376c0450b73cf55bf94",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-17-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/17",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "17",
      "seed": 100000,
      "source_episode_id": "17-grab-roller-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 75
      },
      "steps": 75,
      "simulation_time_s": 26.70000126818195,
      "wall_time_s": 95.564143654,
      "timing": {
        "wall_time_s": 95.564143654,
        "simulation_time_s": 26.70000126818195,
        "physics_steps": 6675,
        "wall_time_start_ns": 1794698094350001194,
        "wall_time_end_ns": 1794698189914157125,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use arms to grab the smooth wooden roller.",
      "native_instruction": "Use arms to grab the smooth wooden roller.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 75 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-17-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.9,
        "native_frames": 76,
        "published_frames": 76,
        "decode_ok": true,
        "sha256": "4320392656b100ed01d7c268ae2d75e8a5bd03f3164202fc4236ff0f0389525e",
        "source_sha256": "e2502428158b909d4974a8ba2acf3a612761fa801d5d1354941a4812c4c4a2f1",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-18-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/18",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "18",
      "seed": 100000,
      "source_episode_id": "18-handover-mic-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 183
      },
      "steps": 183,
      "simulation_time_s": 54.03200256638229,
      "wall_time_s": 370.81270016,
      "timing": {
        "wall_time_s": 370.81270016,
        "simulation_time_s": 54.03200256638229,
        "physics_steps": 13508,
        "wall_time_start_ns": 1794698139420405527,
        "wall_time_end_ns": 1794698510233115914,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use one arm to hold the microphone used for audio input and hand it over.",
      "native_instruction": "Use one arm to hold the microphone used for audio input and hand it over.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 183 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-18-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 6,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 4.6,
        "native_frames": 184,
        "published_frames": 184,
        "decode_ok": true,
        "sha256": "62e5e0cf69bc298e44d7cfd4dc0e8270ca8a7ac8a4e1cb7d4aba78ae5ebc37ec",
        "source_sha256": "50ae6c5f9270626de469926ce44796f0cef5f80643913b556056e38dce71fa53",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-19-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/19",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "19",
      "seed": 100000,
      "source_episode_id": "19-move-can-pot-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 105.59600501554087,
      "wall_time_s": 1226.621328778,
      "timing": {
        "wall_time_s": 1226.621328778,
        "simulation_time_s": 105.59600501554087,
        "physics_steps": 26399,
        "wall_time_start_ns": 1794698218660242084,
        "wall_time_end_ns": 1794699445281574059,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
      "native_instruction": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-19-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "7bc8525f8d900ed1fff9bfafbceee62678ead2307d96732f3b0461e53ad27807",
        "source_sha256": "1219ec3d806f67d09061e6ea95b5ede2ad10fc562b9973d608d94079b7c2c218",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-20-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/20",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "20",
      "seed": 100001,
      "source_episode_id": "20-move-pillbottle-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 115.69200549507514,
      "wall_time_s": 1132.493042413,
      "timing": {
        "wall_time_s": 1132.493042413,
        "simulation_time_s": 115.69200549507514,
        "physics_steps": 28923,
        "wall_time_start_ns": 1794698222000417801,
        "wall_time_end_ns": 1794699354493465610,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the white cylindrical bottle and set it on the pad",
      "native_instruction": "Grab the white cylindrical bottle and set it on the pad",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-20-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "6da853c05dcb69423d366fa37e34db963808213648df641488786af47c187f5f",
        "source_sha256": "51a8b659d9f9e66ac63a210a14388a2b9b8ed17aa29d395a9ca0e38b065c987d",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-21-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/21",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "21",
      "seed": 100000,
      "source_episode_id": "21-move-playingcard-away-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 288
      },
      "steps": 288,
      "simulation_time_s": 80.83600383950397,
      "wall_time_s": 861.566073208,
      "timing": {
        "wall_time_s": 861.566073208,
        "simulation_time_s": 80.83600383950397,
        "physics_steps": 20209,
        "wall_time_start_ns": 1794698261910409241,
        "wall_time_end_ns": 1794699123476487576,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the right arm to grab the hard black box for card deck and relocate it.",
      "native_instruction": "Use the right arm to grab the hard black box for card deck and relocate it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 288 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-21-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 9,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 7.225,
        "native_frames": 289,
        "published_frames": 289,
        "decode_ok": true,
        "sha256": "400d80734ef28b943672e4b1adcbe2eb7cbfa247113f4b9bdfd2ba3d293089fc",
        "source_sha256": "a485f9579b9c54551d3aff757f330f0e0094ffd6007350df856dc1baf4356193",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-22-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/22",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "22",
      "seed": 100000,
      "source_episode_id": "22-move-stapler-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 111.85600531287491,
      "wall_time_s": 1142.340707559,
      "timing": {
        "wall_time_s": 1142.340707559,
        "simulation_time_s": 111.85600531287491,
        "physics_steps": 27964,
        "wall_time_start_ns": 1794698346044230341,
        "wall_time_end_ns": 1794699488384942016,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to move the plastic stapler to Yellow mat",
      "native_instruction": "Use the left arm to move the plastic stapler to Yellow mat",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-22-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "0d00a7e9d51f64acdfc8af8c5f3e7b0d0676f74e46af58468845f4fc5b7be63c",
        "source_sha256": "f1e91a2289da6dc0b6adc752c8a82311f092d1071d0f46d0bf74cfe73d9e82b2",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-23-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/23",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "23",
      "seed": 100000,
      "source_episode_id": "23-open-laptop-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 135.1320064184256,
      "wall_time_s": 2269.670020357,
      "timing": {
        "wall_time_s": 2269.670020357,
        "simulation_time_s": 135.1320064184256,
        "physics_steps": 33783,
        "wall_time_start_ns": 1794698709331529532,
        "wall_time_end_ns": 1794700979001552724,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Open the flat orange laptop with one motion.",
      "native_instruction": "Open the flat orange laptop with one motion.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-23-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "2119f29a359e1321b963512ddee8942b8a661904fa94f1b3a1a8b2f528b64a94",
        "source_sha256": "400aa06ec5545df46304353d65af4ee4bf86a081578f96799560228fabd88456",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-24-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/24",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "24",
      "seed": 100000,
      "source_episode_id": "24-open-microwave-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1500
      },
      "steps": 1500,
      "simulation_time_s": 385.2200182969682,
      "wall_time_s": 4139.833530255,
      "timing": {
        "wall_time_s": 4139.833530255,
        "simulation_time_s": 385.2200182969682,
        "physics_steps": 96305,
        "wall_time_start_ns": 1794698993484474911,
        "wall_time_end_ns": 1794703133318008754,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
      "native_instruction": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-24-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1500,
        "max_control_steps": 1500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 47,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 37.5,
        "native_frames": 1500,
        "published_frames": 1500,
        "decode_ok": true,
        "sha256": "af6b9440bfea4a7dced962fb789adb1fa9e84d89acf8f274092cfc86bb006f6c",
        "source_sha256": "56b62473ab960ad45343feef40c00e744d7f7347805e00a355609f971051f126",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184,
          1216,
          1248,
          1280,
          1312,
          1344,
          1376,
          1408,
          1440,
          1472
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-25-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/25",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "25",
      "seed": 100005,
      "source_episode_id": "25-pick-diverse-bottles-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 82.03200389631093,
      "wall_time_s": 1502.446021056,
      "timing": {
        "wall_time_s": 1502.446021056,
        "simulation_time_s": 82.03200389631093,
        "physics_steps": 20508,
        "wall_time_start_ns": 1794699720371715180,
        "wall_time_end_ns": 1794701222817739893,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
      "native_instruction": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-25-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100005,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "fb7bb5f5023b58d96ed9af2fcce40fd05860efa08f79263b8b0dc153cfd046ba",
        "source_sha256": "ef89311d5a30bdbfd651c57c342565592c827c6541902908527755cdcc3c80ae",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-26-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/26",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "26",
      "seed": 100000,
      "source_episode_id": "26-pick-dual-bottles-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 87.26800414500758,
      "wall_time_s": 1004.596163538,
      "timing": {
        "wall_time_s": 1004.596163538,
        "simulation_time_s": 87.26800414500758,
        "physics_steps": 21817,
        "wall_time_start_ns": 1794699593697185483,
        "wall_time_end_ns": 1794700598293353004,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
      "native_instruction": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-26-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "a259fd8c32277f0d0c39a6610c75e2b417bf25b225cc5db671ede64046240b4d",
        "source_sha256": "3e0b77ccd1204921c5cc009378fb1f60fe1e9e281e1961506b807e8c5ea08c81",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-27-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/27",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "27",
      "seed": 100000,
      "source_episode_id": "27-place-a2b-left-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 124
      },
      "steps": 124,
      "simulation_time_s": 36.676001742016524,
      "wall_time_s": 306.250091339,
      "timing": {
        "wall_time_s": 306.250091339,
        "simulation_time_s": 36.676001742016524,
        "physics_steps": 9169,
        "wall_time_start_ns": 1794699707076632501,
        "wall_time_end_ns": 1794700013326730147,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
      "native_instruction": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 124 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-27-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 3.125,
        "native_frames": 125,
        "published_frames": 125,
        "decode_ok": true,
        "sha256": "8753c09f5b23c1a261aec890a0bd2b4dafc936b8f9d363af77faa978f7a38f39",
        "source_sha256": "f49671e67f1198f08e7c5df401dff9969091144d9b423feef5e7909254b1181d",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-28-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/28",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "28",
      "seed": 100000,
      "source_episode_id": "28-place-a2b-right-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 103.01200489280745,
      "wall_time_s": 1368.564844282,
      "timing": {
        "wall_time_s": 1368.564844282,
        "simulation_time_s": 103.01200489280745,
        "physics_steps": 25753,
        "wall_time_start_ns": 1794700264450369427,
        "wall_time_end_ns": 1794701633015223089,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
      "native_instruction": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-28-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "07502fda83740eeaf28521d3cfe1054e012d2bcc952cf811bafc4b246395fdb1",
        "source_sha256": "d7335622ca55378fa14d7de85496e1933e3df761b1f0e29cf9c9e286e847158a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-29-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/29",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "29",
      "seed": 100000,
      "source_episode_id": "29-place-bread-basket-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 234.1200111201033,
      "wall_time_s": 2107.859850176,
      "timing": {
        "wall_time_s": 2107.859850176,
        "simulation_time_s": 234.1200111201033,
        "physics_steps": 58530,
        "wall_time_start_ns": 1794700562336383922,
        "wall_time_end_ns": 1794702670196238361,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
      "native_instruction": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-29-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "16b0d3dd98e7b5c00a66cca4257074ac257322c7bdc69c9b0a0a8f28a812d142",
        "source_sha256": "c6a74a0be3648c6a8b3c9f47c136de7d9c2b270c214c900d1d09cddad49f2510",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-30-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/30",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "30",
      "seed": 100001,
      "source_episode_id": "30-place-bread-skillet-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 202
      },
      "steps": 202,
      "simulation_time_s": 62.36000296194106,
      "wall_time_s": 680.570418107,
      "timing": {
        "wall_time_s": 680.570418107,
        "simulation_time_s": 62.36000296194106,
        "physics_steps": 15590,
        "wall_time_start_ns": 1794700912877425144,
        "wall_time_end_ns": 1794701593447849726,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
      "native_instruction": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 202 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-30-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 7,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 5.075,
        "native_frames": 203,
        "published_frames": 203,
        "decode_ok": true,
        "sha256": "452b882d2f2f7fd8b594eefac43475f022d7775e5968e60ae0ad90e8f4d30e9e",
        "source_sha256": "760b35507f7aa77e0a9c22e4c7f4b6da33a9bced45fb54f76a8d9e872584b9e4",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-31-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/31",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "31",
      "seed": 100000,
      "source_episode_id": "31-place-burger-fries-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 138.88800659682602,
      "wall_time_s": 1705.095434985,
      "timing": {
        "wall_time_s": 1705.095434985,
        "simulation_time_s": 138.88800659682602,
        "physics_steps": 34722,
        "wall_time_start_ns": 1794701416484875519,
        "wall_time_end_ns": 1794703121580315190,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
      "native_instruction": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-31-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "90018ac4704c5b82afd4ac6ff4690b2defc22d73055ac7803a400e6ff46cc31d",
        "source_sha256": "9d605fd06dcc982e51cfd5a8ef0089e0d62afd5824179cc506f0707561606520",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-32-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/32",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "32",
      "seed": 100000,
      "source_episode_id": "32-place-can-basket-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 240.37601141724735,
      "wall_time_s": 2448.136292759,
      "timing": {
        "wall_time_s": 2448.136292759,
        "simulation_time_s": 240.37601141724735,
        "physics_steps": 60094,
        "wall_time_start_ns": 1794701506672906293,
        "wall_time_end_ns": 1794703954809203697,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "native_instruction": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-32-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "682a81d5762b2bff732757438cdb86618e7eb42835a3396e032cfac93151035a",
        "source_sha256": "5ad4284c73cf8656200ad9d84f38043aaa98ee4e759e42fd5cf8ae168a335ea8",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-33-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/33",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "33",
      "seed": 100000,
      "source_episode_id": "33-place-cans-plasticbox-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 800
      },
      "steps": 800,
      "simulation_time_s": 192.55200914572924,
      "wall_time_s": 2783.626824402,
      "timing": {
        "wall_time_s": 2783.626824402,
        "simulation_time_s": 192.55200914572924,
        "physics_steps": 48138,
        "wall_time_start_ns": 1794701912524306863,
        "wall_time_end_ns": 1794704696151136514,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
      "native_instruction": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 800 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-33-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 25,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 20.0,
        "native_frames": 800,
        "published_frames": 800,
        "decode_ok": true,
        "sha256": "5c51f96c627ded12af27ab914e3d657d1168111d1f6a3983931cba51567bf5a6",
        "source_sha256": "ce444a429a6b6384ace32640bd70de1bad3f6dbfd00231b0d06e28151e98592f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-34-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/34",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "34",
      "seed": 100000,
      "source_episode_id": "34-place-container-plate-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 131
      },
      "steps": 131,
      "simulation_time_s": 49.65200235834345,
      "wall_time_s": 448.187433235,
      "timing": {
        "wall_time_s": 448.187433235,
        "simulation_time_s": 49.65200235834345,
        "physics_steps": 12413,
        "wall_time_start_ns": 1794701972266175278,
        "wall_time_end_ns": 1794702420453615521,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the curved bowl with smooth texture on the gray plate for dining.",
      "native_instruction": "Place the curved bowl with smooth texture on the gray plate for dining.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 131 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-34-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 5,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 3.3,
        "native_frames": 132,
        "published_frames": 132,
        "decode_ok": true,
        "sha256": "179d9eedbdb3f55f78274f1082c1d165b7f436e7a1dda786bd9d5d4695a033f7",
        "source_sha256": "2844b0abf7a06d8ec81eedbabece3d81fee607705a5be98031312d63ccd2074a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-35-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/35",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "35",
      "seed": 100000,
      "source_episode_id": "35-place-empty-cup-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 178
      },
      "steps": 178,
      "simulation_time_s": 58.852002795320004,
      "wall_time_s": 518.547180019,
      "timing": {
        "wall_time_s": 518.547180019,
        "simulation_time_s": 58.852002795320004,
        "physics_steps": 14713,
        "wall_time_start_ns": 1794702708551230676,
        "wall_time_end_ns": 1794703227098418655,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
      "native_instruction": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 178 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-35-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 6,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 4.475,
        "native_frames": 179,
        "published_frames": 179,
        "decode_ok": true,
        "sha256": "b4fef5cfb7cf9847b712b753e6455d6d53834611dd60fc7c226e85d3c09e0dd6",
        "source_sha256": "35f2b9690a00701bd3cfc879e7a419eb42b1b727e18ff642f7e0eec3d192228c",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-36-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/36",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "36",
      "seed": 100000,
      "source_episode_id": "36-place-fan-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 124.98800593661144,
      "wall_time_s": 1240.275089784,
      "timing": {
        "wall_time_s": 1240.275089784,
        "simulation_time_s": 124.98800593661144,
        "physics_steps": 31247,
        "wall_time_start_ns": 1794703075244009388,
        "wall_time_end_ns": 1794704315519103234,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
      "native_instruction": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-36-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "847b5a182178ea9831a3990b9659962291cfe531890d3e703ba59f71cef4a560",
        "source_sha256": "879608ae65b34cd6b572fe4e31f7c0e2a80c60ee60628614aecdc6993d9912e7",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-37-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/37",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "37",
      "seed": 100000,
      "source_episode_id": "37-place-mouse-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 96.42800458008423,
      "wall_time_s": 1015.34059351,
      "timing": {
        "wall_time_s": 1015.34059351,
        "simulation_time_s": 96.42800458008423,
        "physics_steps": 24107,
        "wall_time_start_ns": 1794703403343876771,
        "wall_time_end_ns": 1794704418684474149,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
      "native_instruction": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-37-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "26f4e5b210218b208c5299669e63313a520aa207f1eedb8770e8b8389b908b27",
        "source_sha256": "57187bed5eda016bf36d2cfe7614582428754fb4c186b057748b7e7375d6b4dd",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-38-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/38",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "38",
      "seed": 100000,
      "source_episode_id": "38-place-object-basket-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 182.2120086546056,
      "wall_time_s": 2029.787903367,
      "timing": {
        "wall_time_s": 2029.787903367,
        "simulation_time_s": 182.2120086546056,
        "physics_steps": 45553,
        "wall_time_start_ns": 1794703440346756609,
        "wall_time_end_ns": 1794705470134664009,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
      "native_instruction": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-38-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "6239d0f799c65cf636dbf5a7abaaf919e73bef4d07ab9c247475d1d4eaf219e2",
        "source_sha256": "fbcfb6083e3b430ca128ec75309b63eafbfe48a5cfce22f3af1d2d7dcd6a4e57",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-39-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/39",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "39",
      "seed": 100000,
      "source_episode_id": "39-place-object-scale-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 104.30000495398417,
      "wall_time_s": 1095.802965645,
      "timing": {
        "wall_time_s": 1095.802965645,
        "simulation_time_s": 104.30000495398417,
        "physics_steps": 26075,
        "wall_time_start_ns": 1794704256303510433,
        "wall_time_end_ns": 1794705352106480115,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
      "native_instruction": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-39-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "1166b528a5e2e8626700696f7ac1b9701b3d0a73e2967aaa3088b008e60217df",
        "source_sha256": "5635fb39d6672b36049edf832f7dadec65969654cc9a53e6875e7d493638f010",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-40-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/40",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "40",
      "seed": 100001,
      "source_episode_id": "40-place-object-stand-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 116
      },
      "steps": 116,
      "simulation_time_s": 36.14400171674788,
      "wall_time_s": 306.6485105,
      "timing": {
        "wall_time_s": 306.6485105,
        "simulation_time_s": 36.14400171674788,
        "physics_steps": 9036,
        "wall_time_start_ns": 1794704432545350536,
        "wall_time_end_ns": 1794704739193868979,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the white stapler on the compact black hollow displaystand with the left arm.",
      "native_instruction": "Place the white stapler on the compact black hollow displaystand with the left arm.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 116 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-40-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 2.925,
        "native_frames": 117,
        "published_frames": 117,
        "decode_ok": true,
        "sha256": "1fe137e285f7da6a82159c114912c0a96d541063800016e29e1eed9a20420966",
        "source_sha256": "2d7572d6c8ed004f3a39768cc0eabf834fd2e078082ee5d2798b8d3cc8bbf0a7",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-41-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/41",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "41",
      "seed": 100000,
      "source_episode_id": "41-place-phone-stand-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 134.75600640056655,
      "wall_time_s": 939.976000201,
      "timing": {
        "wall_time_s": 939.976000201,
        "simulation_time_s": 134.75600640056655,
        "physics_steps": 33689,
        "wall_time_start_ns": 1794704338191187074,
        "wall_time_end_ns": 1794705278167190784,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Set the phone with flat black screen on the angled brown phonestand.",
      "native_instruction": "Set the phone with flat black screen on the angled brown phonestand.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-41-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "73dbda50c2e9f459a35c586ad3e2d4d870fb03148b45eb5a72643152839fc7e8",
        "source_sha256": "975bd5ac4267ed82786babedf1c41fac3430ac229de3479865d2caed5d4add3f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-42-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/42",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "42",
      "seed": 100001,
      "source_episode_id": "42-place-shoe-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 128
      },
      "steps": 128,
      "simulation_time_s": 43.12400204828009,
      "wall_time_s": 336.995387519,
      "timing": {
        "wall_time_s": 336.995387519,
        "simulation_time_s": 43.12400204828009,
        "physics_steps": 10781,
        "wall_time_start_ns": 1794704466942210144,
        "wall_time_end_ns": 1794704803937608519,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the medium adult shoe from the table and set it on the mat",
      "native_instruction": "Pick up the medium adult shoe from the table and set it on the mat",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 128 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-42-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 3.225,
        "native_frames": 129,
        "published_frames": 129,
        "decode_ok": true,
        "sha256": "27c39825be8b848d94cf813ee20da38c5719370f9748c0892acadc89118aa0cc",
        "source_sha256": "1e75586bbb019b9f4b8b7547cae2c5291553aa3f0c538e03cc7adfa2e3b2a131",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-43-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/43",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "43",
      "seed": 100000,
      "source_episode_id": "43-press-stapler-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 189
      },
      "steps": 189,
      "simulation_time_s": 41.356001964304596,
      "wall_time_s": 555.560583846,
      "timing": {
        "wall_time_s": 555.560583846,
        "simulation_time_s": 41.356001964304596,
        "physics_steps": 10339,
        "wall_time_start_ns": 1794704452780488834,
        "wall_time_end_ns": 1794705008341078100,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Press the plastic stapler firmly using your arm.",
      "native_instruction": "Press the plastic stapler firmly using your arm.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 189 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-43-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 6,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 4.75,
        "native_frames": 190,
        "published_frames": 190,
        "decode_ok": true,
        "sha256": "52470ecc6534c3fecf5a5f2b25bea94ea0191ed1fc155eaf74fa0faef88bee6b",
        "source_sha256": "b1217f265e80229b67cf37ac52c8139940c93743acf92fc0429cebeee99fd078",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-44-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/44",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "44",
      "seed": 100000,
      "source_episode_id": "44-rotate-qrcode-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 121.18000575574115,
      "wall_time_s": 909.818378811,
      "timing": {
        "wall_time_s": 909.818378811,
        "simulation_time_s": 121.18000575574115,
        "physics_steps": 30295,
        "wall_time_start_ns": 1794704524571285505,
        "wall_time_end_ns": 1794705434389670938,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
      "native_instruction": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-44-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "c8de0edb171d792af86c68bae4af81db14646dff648c93e0aac486ec1e97a967",
        "source_sha256": "688a0d9254d29b6e6d3b772d4e43fed85dbff576bff13584f2a7a416155adec3",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-45-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/45",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "45",
      "seed": 100001,
      "source_episode_id": "45-shake-bottle-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 77
      },
      "steps": 77,
      "simulation_time_s": 25.43600120814517,
      "wall_time_s": 322.543041113,
      "timing": {
        "wall_time_s": 322.543041113,
        "simulation_time_s": 25.43600120814517,
        "physics_steps": 6359,
        "wall_time_start_ns": 1794704619150906961,
        "wall_time_end_ns": 1794704941693953105,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the smooth bottle with blue text and shake it.",
      "native_instruction": "Pick up the smooth bottle with blue text and shake it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 77 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-45-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.95,
        "native_frames": 78,
        "published_frames": 78,
        "decode_ok": true,
        "sha256": "1489c0f5fd08e47212f452e287c1c2ecb34022ebf48f0cf9ac02b49fcc81c4f1",
        "source_sha256": "0f8a61e8217c2d41e39c1df728b3f349e695ff9aaa8fc4b3100fcf4416f7a483",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-46-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/46",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "46",
      "seed": 100001,
      "source_episode_id": "46-shake-bottle-horizontally-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 79
      },
      "steps": 79,
      "simulation_time_s": 25.980001233983785,
      "wall_time_s": 373.296846288,
      "timing": {
        "wall_time_s": 373.296846288,
        "simulation_time_s": 25.980001233983785,
        "physics_steps": 6495,
        "wall_time_start_ns": 1794704635380139368,
        "wall_time_end_ns": 1794705008676995722,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
      "native_instruction": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 79 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-46-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 2.0,
        "native_frames": 80,
        "published_frames": 80,
        "decode_ok": true,
        "sha256": "c19eab298371a51ba8961cda7a4806ad05774e5daa212bf88088fd4cc267a166",
        "source_sha256": "73c7edbf2dafccebe774be3bfd12d1669d61ef824e42e1d4736c082b6db7cf97",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-47-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/47",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "47",
      "seed": 100000,
      "source_episode_id": "47-stack-blocks-two-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 274
      },
      "steps": 274,
      "simulation_time_s": 84.42000400973484,
      "wall_time_s": 751.612961199,
      "timing": {
        "wall_time_s": 751.612961199,
        "simulation_time_s": 84.42000400973484,
        "physics_steps": 21105,
        "wall_time_start_ns": 1794704788399198710,
        "wall_time_end_ns": 1794705540012165019,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Move red block to the center, then put green block on top.",
      "native_instruction": "Move red block to the center, then put green block on top.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 274 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-47-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 9,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 6.875,
        "native_frames": 275,
        "published_frames": 275,
        "decode_ok": true,
        "sha256": "0e0464a3353afc02b343cfbcafbabe7d4006a7cd277b4b85a00f98b3460b5d6a",
        "source_sha256": "43c84ba43df72a2b172f2d763534e30ac65e00fdde0b4fc5856869ee284cdbc6",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-48-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/48",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "48",
      "seed": 100000,
      "source_episode_id": "48-stack-bowls-two-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 313
      },
      "steps": 313,
      "simulation_time_s": 94.41600448451936,
      "wall_time_s": 788.251499084,
      "timing": {
        "wall_time_s": 788.251499084,
        "simulation_time_s": 94.41600448451936,
        "physics_steps": 23604,
        "wall_time_start_ns": 1794704901618321474,
        "wall_time_end_ns": 1794705689869826515,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
      "native_instruction": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 313 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-48-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 10,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 7.85,
        "native_frames": 314,
        "published_frames": 314,
        "decode_ok": true,
        "sha256": "84c7f2625aab246ee8bde0841a6ca0369868a758f1dd1cb268df46a217e30064",
        "source_sha256": "2c5bb9c3dca1a820a50099c44959ee6bebea70e545a55444c653c3586351d9af",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-49-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/49",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "49",
      "seed": 100003,
      "source_episode_id": "49-stamp-seal-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 125.53200596245006,
      "wall_time_s": 768.810585809,
      "timing": {
        "wall_time_s": 768.810585809,
        "simulation_time_s": 125.53200596245006,
        "physics_steps": 31383,
        "wall_time_start_ns": 1794704996066030120,
        "wall_time_end_ns": 1794705764876619483,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the block-shaped seal with angled edges, press onto Blue",
      "native_instruction": "Grab the block-shaped seal with angled edges, press onto Blue",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-49-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100003,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "9422eb7aba9e79a22fb7cbceba974771480f1ef2dda67c2ee32cc43840cbcceb",
        "source_sha256": "b11640ece9ce6a438cd1703eeb7e3c8341acd74fad39b9bb7acfd63783d00200",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-image_only-50-20261113",
      "variant": "jy-mode1",
      "task_key": "task03/50",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "50",
      "seed": 100000,
      "source_episode_id": "50-turn-switch-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 94.82000450370833,
      "wall_time_s": 764.82104345,
      "timing": {
        "wall_time_s": 764.82104345,
        "simulation_time_s": 94.82000450370833,
        "physics_steps": 23705,
        "wall_time_start_ns": 1794705086093217353,
        "wall_time_end_ns": 1794705850914266600,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Press the flat square gray switch using the left arm",
      "native_instruction": "Press the flat square gray switch using the left arm",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-image_only-50-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "3302a6539666a6b6c49f2086d1708ebb1c9dd08ec2b682434dade252372aa36e",
        "source_sha256": "1f5d59263dd8fd19eef9a09a45b05f8406dba1a8d884349d46f5604c68308cfe",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-01-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/01",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "01",
      "seed": 100001,
      "source_episode_id": "01-handover-block-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 800
      },
      "steps": 800,
      "simulation_time_s": 202.18800960341468,
      "wall_time_s": 2437.401726425,
      "timing": {
        "wall_time_s": 2437.401726425,
        "simulation_time_s": 202.18800960341468,
        "physics_steps": 50547,
        "wall_time_start_ns": 1794687770290306836,
        "wall_time_end_ns": 1794690207692037855,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
      "native_instruction": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 800 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-01-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 25,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 20.0,
        "native_frames": 800,
        "published_frames": 800,
        "decode_ok": true,
        "sha256": "0c2b41dce2c5861c66ccb6653bf5757346a29f33830201ce001c24e1ad5210a1",
        "source_sha256": "d433de062e87a6b344d9b3e2b66c263ecc532303af9f50273a83fc8c5fa93f1b",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-02-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/02",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "02",
      "seed": 100001,
      "source_episode_id": "02-place-dual-shoes-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 600
      },
      "steps": 600,
      "simulation_time_s": 221.4880105201155,
      "wall_time_s": 1729.046285819,
      "timing": {
        "wall_time_s": 1729.046285819,
        "simulation_time_s": 221.4880105201155,
        "physics_steps": 55372,
        "wall_time_start_ns": 1794687793306095814,
        "wall_time_end_ns": 1794689522352385591,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
      "native_instruction": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 600 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-02-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 15.0,
        "native_frames": 600,
        "published_frames": 600,
        "decode_ok": true,
        "sha256": "e39b35b5c0cf32019e703259e2951456485639dd5c6edb59cea03b1e50de305e",
        "source_sha256": "0e31df1e04fdc0e1cc3e5bc52439bc03ca079479c83d60e118ebe36376df071e",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-03-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/03",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "03",
      "seed": 100000,
      "source_episode_id": "03-put-bottles-dustbin-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1700
      },
      "steps": 1700,
      "simulation_time_s": 524.8880249308422,
      "wall_time_s": 3782.484527524,
      "timing": {
        "wall_time_s": 3782.484527524,
        "simulation_time_s": 524.8880249308422,
        "physics_steps": 131222,
        "wall_time_start_ns": 1794687835465719595,
        "wall_time_end_ns": 1794691617950250486,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
      "native_instruction": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-03-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1700,
        "max_control_steps": 1700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 54,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 42.5,
        "native_frames": 1700,
        "published_frames": 1700,
        "decode_ok": true,
        "sha256": "145aa52d54a0311b351695be6e49ffc18eadedba0cc26f24fdec5c2d7cf90c44",
        "source_sha256": "98946b72f28238165692a61f3ee574c47094d65dc8c06d82423de6e879849682",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184,
          1216,
          1248,
          1280,
          1312,
          1344,
          1376,
          1408,
          1440,
          1472,
          1504,
          1536,
          1568,
          1600,
          1632,
          1664,
          1696
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-04-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/04",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "04",
      "seed": 100001,
      "source_episode_id": "04-scan-object-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 159
      },
      "steps": 159,
      "simulation_time_s": 56.86000270070508,
      "wall_time_s": 366.52074051,
      "timing": {
        "wall_time_s": 366.52074051,
        "simulation_time_s": 56.86000270070508,
        "physics_steps": 14215,
        "wall_time_start_ns": 1794605921870596744,
        "wall_time_end_ns": 1794606288391351317,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
      "native_instruction": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 159 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-04-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 5,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 4.0,
        "native_frames": 160,
        "published_frames": 160,
        "decode_ok": true,
        "sha256": "cb6d6ab30bfbc10441e49bf04f7786cacdf274fbba79dcbdf6b80847764aafbc",
        "source_sha256": "5007d3570bd73c12b19889d3408fcb68af8e3a6d2bc5e6ed765fd61afc6377b2",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-05-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/05",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "05",
      "seed": 100000,
      "source_episode_id": "05-stack-bowls-three-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 573
      },
      "steps": 573,
      "simulation_time_s": 173.59200824517757,
      "wall_time_s": 1819.185870494,
      "timing": {
        "wall_time_s": 1819.185870494,
        "simulation_time_s": 173.59200824517757,
        "physics_steps": 43398,
        "wall_time_start_ns": 1794687888966837060,
        "wall_time_end_ns": 1794689708152713289,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
      "native_instruction": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 573 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-05-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 18,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 14.35,
        "native_frames": 574,
        "published_frames": 574,
        "decode_ok": true,
        "sha256": "47c0798a880563f89e4c061e53616f8ca9abaed597cd3020b272d6a03bebb4c9",
        "source_sha256": "9e4b820df23eb4a5f8aa3a2db066ae5259a27f4dc2ccaebffa7d8b5c33739e79",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-06-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/06",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "06",
      "seed": 100000,
      "source_episode_id": "06-stack-blocks-three-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 324.30001540342346,
      "wall_time_s": 3477.132275232,
      "timing": {
        "wall_time_s": 3477.132275232,
        "simulation_time_s": 324.30001540342346,
        "physics_steps": 81075,
        "wall_time_start_ns": 1794687956893680643,
        "wall_time_end_ns": 1794691434025958461,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
      "native_instruction": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-06-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "5ce64e9d1805a1db290a2ed3ee1b02101b5630f792a0404d7794ffcc03fd58e0",
        "source_sha256": "3a94c48c76ba78990af50677b41dee5cbf0be4f185d4febc90f0a074cb3b5dcd",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-07-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/07",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "07",
      "seed": 100000,
      "source_episode_id": "07-hanging-mug-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 900
      },
      "steps": 900,
      "simulation_time_s": 238.5160113289021,
      "wall_time_s": 3271.41139538,
      "timing": {
        "wall_time_s": 3271.41139538,
        "simulation_time_s": 238.5160113289021,
        "physics_steps": 59629,
        "wall_time_start_ns": 1794688035146342974,
        "wall_time_end_ns": 1794691306557742934,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
      "native_instruction": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 900 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-07-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 29,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 22.5,
        "native_frames": 900,
        "published_frames": 900,
        "decode_ok": true,
        "sha256": "bc649012d4884a0a3c801ea91f0cdd548a78050e048c3ab7bf08be18e6322929",
        "source_sha256": "50e4f04ab968e66afe26e1a18596d541d8d689146394248566a19257f6bc0a64",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-08-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/08",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "08",
      "seed": 100001,
      "source_episode_id": "08-put-object-cabinet-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 147.00400698231533,
      "wall_time_s": 2637.237403735,
      "timing": {
        "wall_time_s": 2637.237403735,
        "simulation_time_s": 147.00400698231533,
        "physics_steps": 36751,
        "wall_time_start_ns": 1794688091420572072,
        "wall_time_end_ns": 1794690728657980127,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
      "native_instruction": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-08-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "47c2ff7a8edb32c7da6407ed73252ce7eb4f5dcb44ed6898b946305a0c27c4e6",
        "source_sha256": "648da914182fc96cb41aba805d5b0221d81185853fc782b15edec4bce027bfbb",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-09-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/09",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "09",
      "seed": 100000,
      "source_episode_id": "09-blocks-ranking-size-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 429
      },
      "steps": 429,
      "simulation_time_s": 126.38800600310788,
      "wall_time_s": 1368.365401642,
      "timing": {
        "wall_time_s": 1368.365401642,
        "simulation_time_s": 126.38800600310788,
        "physics_steps": 31597,
        "wall_time_start_ns": 1794689839097834273,
        "wall_time_end_ns": 1794691207463246661,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place large block, medium block, and small block in order at the table's center.",
      "native_instruction": "Place large block, medium block, and small block in order at the table's center.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 429 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-09-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 14,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.75,
        "native_frames": 430,
        "published_frames": 430,
        "decode_ok": true,
        "sha256": "d03b93c78de30f6369575f8c8635a364a1ff5cd851ab3a1a6cec32a25f29dcd5",
        "source_sha256": "30e9d9f80bf6e5c3904d5c9f78a80d6b9fb9431902f446971ddaa1541b9966da",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-10-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/10",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "10",
      "seed": 100000,
      "source_episode_id": "10-lift-pot-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 73.46000348916277,
      "wall_time_s": 1320.283051355,
      "timing": {
        "wall_time_s": 1320.283051355,
        "simulation_time_s": 73.46000348916277,
        "physics_steps": 18365,
        "wall_time_start_ns": 1794689955689696582,
        "wall_time_end_ns": 1794691275972751547,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the dark gray cooking kitchenpot using both arms",
      "native_instruction": "Lift the dark gray cooking kitchenpot using both arms",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-10-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "58c55c708c48ed5a340a25cb767a3844db6df954c1bd8189060700a8df88d08c",
        "source_sha256": "ca1329e21d634ad804df8b97716ce9dab4ad242b8792c8b97aaac870a01c90c2",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-11-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/11",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "11",
      "seed": 100000,
      "source_episode_id": "11-adjust-bottle-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 97.696004640311,
      "wall_time_s": 1399.135856697,
      "timing": {
        "wall_time_s": 1399.135856697,
        "simulation_time_s": 97.696004640311,
        "physics_steps": 24424,
        "wall_time_start_ns": 1794690560195706525,
        "wall_time_end_ns": 1794691959331567051,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the brown bottle upright",
      "native_instruction": "Use the left arm to grab the brown bottle upright",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-11-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "019325c8622b6586c23a8b1fd3915ce38e02a40775d0e1080bc5555b05e63f54",
        "source_sha256": "5c13e82b6728409fd4adddb18c49d5d9207193518864a26e4cebf36039dbd5fc",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-12-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/12",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "12",
      "seed": 100000,
      "source_episode_id": "12-beat-block-hammer-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 85.08800404146314,
      "wall_time_s": 694.184199098,
      "timing": {
        "wall_time_s": 694.184199098,
        "simulation_time_s": 85.08800404146314,
        "physics_steps": 21272,
        "wall_time_start_ns": 1794690269305395672,
        "wall_time_end_ns": 1794690963489598836,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the hammer with black handle then beat block",
      "native_instruction": "Use the left arm to grab the hammer with black handle then beat block",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-12-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "309999567bc57a5f577c7ac4b5b142a1b80279bd8f40588340c6b711acf4c0d0",
        "source_sha256": "821a7ec93c9239786bee87eac20fab0cbf142b2400429f11545e0d3396c865bf",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-13-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/13",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "13",
      "seed": 100000,
      "source_episode_id": "13-blocks-ranking-rgb-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 387.4640184035525,
      "wall_time_s": 1833.015227245,
      "timing": {
        "wall_time_s": 1833.015227245,
        "simulation_time_s": 387.4640184035525,
        "physics_steps": 96866,
        "wall_time_start_ns": 1794690284060693218,
        "wall_time_end_ns": 1794692117075923846,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
      "native_instruction": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-13-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "5c51e3d6cdd6b0bfc0ccfccaa9eb381b61958b6672509ac8af2904d538fd4293",
        "source_sha256": "8463e035ae576e8d7b0961305a04c1dc08e8a96475977cd41630e428c9883996",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-14-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/14",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "14",
      "seed": 100000,
      "source_episode_id": "14-click-alarmclock-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 64
      },
      "steps": 64,
      "simulation_time_s": 21.56400102423504,
      "wall_time_s": 123.490053314,
      "timing": {
        "wall_time_s": 123.490053314,
        "simulation_time_s": 21.56400102423504,
        "physics_steps": 5391,
        "wall_time_start_ns": 1794690333074578403,
        "wall_time_end_ns": 1794690456564638190,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Tap the center top button of the alarm-clock with beige screen",
      "native_instruction": "Tap the center top button of the alarm-clock with beige screen",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 64 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-14-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 2,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.625,
        "native_frames": 65,
        "published_frames": 65,
        "decode_ok": true,
        "sha256": "e81aa1a1e6663f6d519beb20f23022607d586f16d161185510ab83c501f4641e",
        "source_sha256": "597eb8a3e9dc436e51b5ed67044bd451f55bc66f13d311d402dc502072d3a91c",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-15-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/15",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "15",
      "seed": 100000,
      "source_episode_id": "15-click-bell-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 59
      },
      "steps": 59,
      "simulation_time_s": 23.276001105550677,
      "wall_time_s": 164.759319445,
      "timing": {
        "wall_time_s": 164.759319445,
        "simulation_time_s": 23.276001105550677,
        "physics_steps": 5819,
        "wall_time_start_ns": 1794690411443981828,
        "wall_time_end_ns": 1794690576203306725,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to press the medium-sized bell's top center.",
      "native_instruction": "Use the left arm to press the medium-sized bell's top center.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 59 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-15-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 2,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.5,
        "native_frames": 60,
        "published_frames": 60,
        "decode_ok": true,
        "sha256": "9c4ef952d9acfd1c4e1ed1d260f0e10f4991d28c50d943748d821368bd124f2f",
        "source_sha256": "84a42fac6c30bf1115d34da71a1370acfa96d66231f1bdb037c02c6d1fad0430",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-16-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/16",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "16",
      "seed": 100006,
      "source_episode_id": "16-dump-bin-bigbin-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 321
      },
      "steps": 321,
      "simulation_time_s": 103.86000493308529,
      "wall_time_s": 755.79046931,
      "timing": {
        "wall_time_s": 755.79046931,
        "simulation_time_s": 103.86000493308529,
        "physics_steps": 25965,
        "wall_time_start_ns": 1794690567670664163,
        "wall_time_end_ns": 1794691323461140350,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
      "native_instruction": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 321 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-16-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100006,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 11,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 8.05,
        "native_frames": 322,
        "published_frames": 322,
        "decode_ok": true,
        "sha256": "202406778ffb9c7f0d5f3b334277f5586201ea50a74ce2713136a5dffc79b26b",
        "source_sha256": "9b72586614a60dee9f98501f8e892b4d036029c8d42335aaeac09391e75ff298",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-17-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/17",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "17",
      "seed": 100000,
      "source_episode_id": "17-grab-roller-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 76
      },
      "steps": 76,
      "simulation_time_s": 27.464001304470003,
      "wall_time_s": 123.844073475,
      "timing": {
        "wall_time_s": 123.844073475,
        "simulation_time_s": 27.464001304470003,
        "physics_steps": 6866,
        "wall_time_start_ns": 1794698099551625245,
        "wall_time_end_ns": 1794698223395706826,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use arms to grab the smooth wooden roller.",
      "native_instruction": "Use arms to grab the smooth wooden roller.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 76 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-17-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.925,
        "native_frames": 77,
        "published_frames": 77,
        "decode_ok": true,
        "sha256": "fc017ebea05e41a938d8902eb401f8a7c99d812e6a07a39e5155001076684e73",
        "source_sha256": "d5986884f063d905f027d34578a7910912cdee9ff270320b9e25b5ecae5d5626",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-18-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/18",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "18",
      "seed": 100000,
      "source_episode_id": "18-handover-mic-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 204
      },
      "steps": 204,
      "simulation_time_s": 57.54800273338333,
      "wall_time_s": 452.198213127,
      "timing": {
        "wall_time_s": 452.198213127,
        "simulation_time_s": 57.54800273338333,
        "physics_steps": 14387,
        "wall_time_start_ns": 1794698133639598049,
        "wall_time_end_ns": 1794698585837824792,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use one arm to hold the microphone used for audio input and hand it over.",
      "native_instruction": "Use one arm to hold the microphone used for audio input and hand it over.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 204 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-18-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 7,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 5.125,
        "native_frames": 205,
        "published_frames": 205,
        "decode_ok": true,
        "sha256": "74e2701e693b3657d5fd496e70b75bbed725b8546c94d4eb7338dcde467aa612",
        "source_sha256": "3b422b985fdd711fed704fd1ccf7391e6fefd751ce29d80cc6a61e6ab2450cc3",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-19-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/19",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "19",
      "seed": 100000,
      "source_episode_id": "19-move-can-pot-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 100.46400477178395,
      "wall_time_s": 1081.655860749,
      "timing": {
        "wall_time_s": 1081.655860749,
        "simulation_time_s": 100.46400477178395,
        "physics_steps": 25116,
        "wall_time_start_ns": 1794698197798263183,
        "wall_time_end_ns": 1794699279454127629,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
      "native_instruction": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-19-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "40dd1b7f63ba8500bc65a0a52354ea36befcecbd1ba727ca7b6150d9537cbe80",
        "source_sha256": "479837166b32da7cba6672bc8ac186aefe46b52bdf16049f2524be2393a886ba",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-20-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/20",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "20",
      "seed": 100001,
      "source_episode_id": "20-move-pillbottle-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 114.84800545498729,
      "wall_time_s": 1287.502476263,
      "timing": {
        "wall_time_s": 1287.502476263,
        "simulation_time_s": 114.84800545498729,
        "physics_steps": 28712,
        "wall_time_start_ns": 1794698295357470927,
        "wall_time_end_ns": 1794699582859956805,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the white cylindrical bottle and set it on the pad",
      "native_instruction": "Grab the white cylindrical bottle and set it on the pad",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-20-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "2b83f0bd6aecf9dff013c42fd04b1de9750215c65528e1c79ac32b2b7b62dbba",
        "source_sha256": "94551f575c33ce1c2f952b60d6484b37b67520fbb16f812919b96342cfb7dc3b",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-21-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/21",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "21",
      "seed": 100000,
      "source_episode_id": "21-move-playingcard-away-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 190
      },
      "steps": 190,
      "simulation_time_s": 53.94400256220251,
      "wall_time_s": 602.390783569,
      "timing": {
        "wall_time_s": 602.390783569,
        "simulation_time_s": 53.94400256220251,
        "physics_steps": 13486,
        "wall_time_start_ns": 1794698305088585188,
        "wall_time_end_ns": 1794698907479375113,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the right arm to grab the hard black box for card deck and relocate it.",
      "native_instruction": "Use the right arm to grab the hard black box for card deck and relocate it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 190 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-21-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 6,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 4.775,
        "native_frames": 191,
        "published_frames": 191,
        "decode_ok": true,
        "sha256": "e575cfe484a628b880fea3384c768c09bb0f2df1f4ac6720382da429c35b841f",
        "source_sha256": "c815c8eceb4e7ce73582cd277cbe94149ef2b0ef42a6bb661a04539afedb4a2d",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-22-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/22",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "22",
      "seed": 100000,
      "source_episode_id": "22-move-stapler-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 96.07200456317514,
      "wall_time_s": 1162.570869347,
      "timing": {
        "wall_time_s": 1162.570869347,
        "simulation_time_s": 96.07200456317514,
        "physics_steps": 24018,
        "wall_time_start_ns": 1794698572682832330,
        "wall_time_end_ns": 1794699735253705538,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to move the plastic stapler to Yellow mat",
      "native_instruction": "Use the left arm to move the plastic stapler to Yellow mat",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-22-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "44a8ac26ba140a00f6a6fd38a490d15c05aa1c1cd3bd99c7ffaf1ab5c7a0b5fc",
        "source_sha256": "5bd6e1076420ea3a94544c0c48f2af2df419775d3ffe361ec6cb48524e8aeb43",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-23-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/23",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "23",
      "seed": 100000,
      "source_episode_id": "23-open-laptop-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 131.56800624914467,
      "wall_time_s": 2277.301896476,
      "timing": {
        "wall_time_s": 2277.301896476,
        "simulation_time_s": 131.56800624914467,
        "physics_steps": 32892,
        "wall_time_start_ns": 1794698900099606150,
        "wall_time_end_ns": 1794701177401506402,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Open the flat orange laptop with one motion.",
      "native_instruction": "Open the flat orange laptop with one motion.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-23-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "7e7590df5b68bec0ed7d8178f8a49d5f9cc0cf8d5834faa82aa3cac698b2f139",
        "source_sha256": "2974debb4fb7ebe33207ab9fe0c6bb8cf114d1789c7ac33152a93ca3c22be4b1",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-24-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/24",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "24",
      "seed": 100000,
      "source_episode_id": "24-open-microwave-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1500
      },
      "steps": 1500,
      "simulation_time_s": 389.59201850462705,
      "wall_time_s": 4806.330562381,
      "timing": {
        "wall_time_s": 4806.330562381,
        "simulation_time_s": 389.59201850462705,
        "physics_steps": 97398,
        "wall_time_start_ns": 1794699277730551000,
        "wall_time_end_ns": 1794704084061123751,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
      "native_instruction": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-24-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1500,
        "max_control_steps": 1500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 47,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 37.5,
        "native_frames": 1500,
        "published_frames": 1500,
        "decode_ok": true,
        "sha256": "5a211e82ae51b9b5b8df6fb59fafc0745f84fbbf856680e5526dab4198a65e1f",
        "source_sha256": "ca80d2d398217b8bfdf6c8086df6b41108565037e14a5cda68ac752bc47b4609",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184,
          1216,
          1248,
          1280,
          1312,
          1344,
          1376,
          1408,
          1440,
          1472
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-25-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/25",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "25",
      "seed": 100005,
      "source_episode_id": "25-pick-diverse-bottles-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 79.41600377205759,
      "wall_time_s": 808.871778024,
      "timing": {
        "wall_time_s": 808.871778024,
        "simulation_time_s": 79.41600377205759,
        "physics_steps": 19854,
        "wall_time_start_ns": 1794699379967069431,
        "wall_time_end_ns": 1794700188838851493,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
      "native_instruction": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-25-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100005,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "07f00d91e140f934503eb02247865f8d6edbb5771fe34fefe5d2d419daf60db0",
        "source_sha256": "d9e0bb40c9398142670ed55cb992a91042efbb70f8d4389a21f5ed8add6f8d06",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-26-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/26",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "26",
      "seed": 100000,
      "source_episode_id": "26-pick-dual-bottles-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 84.80400402797386,
      "wall_time_s": 935.262900454,
      "timing": {
        "wall_time_s": 935.262900454,
        "simulation_time_s": 84.80400402797386,
        "physics_steps": 21201,
        "wall_time_start_ns": 1794699669294099831,
        "wall_time_end_ns": 1794700604557004189,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
      "native_instruction": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-26-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "1e06e440cb38aeabb39a94f74bdd5ef5b820f352d7bc00e9908f355a0db6397f",
        "source_sha256": "29bd00abb821a7f434c10eac548809e197944d8f82294d4f37597fdc98d20f1a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-27-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/27",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "27",
      "seed": 100000,
      "source_episode_id": "27-place-a2b-left-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 125
      },
      "steps": 125,
      "simulation_time_s": 37.1160017629154,
      "wall_time_s": 425.723152596,
      "timing": {
        "wall_time_s": 425.723152596,
        "simulation_time_s": 37.1160017629154,
        "physics_steps": 9279,
        "wall_time_start_ns": 1794700126459545783,
        "wall_time_end_ns": 1794700552182703566,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
      "native_instruction": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 125 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-27-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 3.15,
        "native_frames": 126,
        "published_frames": 126,
        "decode_ok": true,
        "sha256": "70abaef90a42518f6bd95584a5828cb4cb4400fa0be629437a633bd6da97d034",
        "source_sha256": "68660df90e29c79bd8f998086fafb4c0d3c5a96cd3f3cf8a908953789d615bf1",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-28-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/28",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "28",
      "seed": 100000,
      "source_episode_id": "28-place-a2b-right-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 102.02800484606996,
      "wall_time_s": 1128.368083281,
      "timing": {
        "wall_time_s": 1128.368083281,
        "simulation_time_s": 102.02800484606996,
        "physics_steps": 25507,
        "wall_time_start_ns": 1794700448460361797,
        "wall_time_end_ns": 1794701576828448587,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
      "native_instruction": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-28-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "237b25dd70660790767ba0f03e3d82f13392e75bf81c442ba31f20fd7cf9747e",
        "source_sha256": "bfb1dd9e97c60acf17e73a86139ef08c42c6623364870006706b9ceab26d3248",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-29-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/29",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "29",
      "seed": 100000,
      "source_episode_id": "29-place-bread-basket-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 385
      },
      "steps": 385,
      "simulation_time_s": 136.57200648682192,
      "wall_time_s": 1101.447539739,
      "timing": {
        "wall_time_s": 1101.447539739,
        "simulation_time_s": 136.57200648682192,
        "physics_steps": 34143,
        "wall_time_start_ns": 1794700812555265908,
        "wall_time_end_ns": 1794701914002811371,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
      "native_instruction": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 385 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-29-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 9.65,
        "native_frames": 386,
        "published_frames": 386,
        "decode_ok": true,
        "sha256": "88f532918c2c94a9590e8ab814b692b170d22bc0890be77e83ef70a757c201ff",
        "source_sha256": "8f85ea922bfb4adce4aeb6d801b4f0cce6fbb2dcbf1d4fc6e0f42a70505a6b6f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-30-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/30",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "30",
      "seed": 100001,
      "source_episode_id": "30-place-bread-skillet-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 204
      },
      "steps": 204,
      "simulation_time_s": 61.7560029332526,
      "wall_time_s": 655.069479997,
      "timing": {
        "wall_time_s": 655.069479997,
        "simulation_time_s": 61.7560029332526,
        "physics_steps": 15439,
        "wall_time_start_ns": 1794700875214458423,
        "wall_time_end_ns": 1794701530283944137,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
      "native_instruction": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 204 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-30-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 7,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 5.125,
        "native_frames": 205,
        "published_frames": 205,
        "decode_ok": true,
        "sha256": "68bdfee3815aea9c4fb5ca7221bcda0cdfd2c492753eef8b137595c644acc880",
        "source_sha256": "be49377cfad93f57dac19f08913646719f4a9c309d6fbff1d276beb13ca2d75a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-31-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/31",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "31",
      "seed": 100000,
      "source_episode_id": "31-place-burger-fries-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 148.95600707503036,
      "wall_time_s": 1416.10027606,
      "timing": {
        "wall_time_s": 1416.10027606,
        "simulation_time_s": 148.95600707503036,
        "physics_steps": 37239,
        "wall_time_start_ns": 1794700925129023412,
        "wall_time_end_ns": 1794702341229303604,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
      "native_instruction": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-31-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "63d511d211cbf139fb3a27994123d7dd1429334e69e7f893a81128a6f01beab3",
        "source_sha256": "b9bc7e5016cb1084963fbe9e6f6834415dbbee6da8ca13470c6c055a35ce387a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-32-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/32",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "32",
      "seed": 100000,
      "source_episode_id": "32-place-can-basket-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 220.21601045969874,
      "wall_time_s": 2257.184771613,
      "timing": {
        "wall_time_s": 2257.184771613,
        "simulation_time_s": 220.21601045969874,
        "physics_steps": 55054,
        "wall_time_start_ns": 1794702021892976809,
        "wall_time_end_ns": 1794704279077752454,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "native_instruction": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-32-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "a2d259d6744dac204bc071d81cf37865c88545873285afcd3f75271ecf15f75c",
        "source_sha256": "5a1f255734cb9ba6f9f4062a18854d940aee735420215995f840840fc0dc9928",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-33-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/33",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "33",
      "seed": 100000,
      "source_episode_id": "33-place-cans-plasticbox-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 800
      },
      "steps": 800,
      "simulation_time_s": 235.00401116209105,
      "wall_time_s": 2751.55804251,
      "timing": {
        "wall_time_s": 2751.55804251,
        "simulation_time_s": 235.00401116209105,
        "physics_steps": 58751,
        "wall_time_start_ns": 1794701861875977271,
        "wall_time_end_ns": 1794704613434024603,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
      "native_instruction": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 800 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-33-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 25,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 20.0,
        "native_frames": 800,
        "published_frames": 800,
        "decode_ok": true,
        "sha256": "15b99081c7fb0d0d2d70417def5e5beaeb0bbcfde57ca78d85678d3b26400b4a",
        "source_sha256": "cdfa875d974a96f81cd2fb42940f3cfa13b2f7b0abb0a0a5f2a791b702e3b002",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-34-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/34",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "34",
      "seed": 100000,
      "source_episode_id": "34-place-container-plate-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 129
      },
      "steps": 129,
      "simulation_time_s": 48.73600231483579,
      "wall_time_s": 394.034290891,
      "timing": {
        "wall_time_s": 394.034290891,
        "simulation_time_s": 48.73600231483579,
        "physics_steps": 12184,
        "wall_time_start_ns": 1794702348003549798,
        "wall_time_end_ns": 1794702742037845941,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the curved bowl with smooth texture on the gray plate for dining.",
      "native_instruction": "Place the curved bowl with smooth texture on the gray plate for dining.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 129 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-34-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 5,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 3.25,
        "native_frames": 130,
        "published_frames": 130,
        "decode_ok": true,
        "sha256": "aff578e27c14b61ccaf231dfe9eb5d48e847361b573cf81e671ee2863d8956ea",
        "source_sha256": "0e631b8bec4a4f8a859e63e1f6b3154edf82df1da975e2d025c9aa954cd22e81",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-35-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/35",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "35",
      "seed": 100000,
      "source_episode_id": "35-place-empty-cup-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 156
      },
      "steps": 156,
      "simulation_time_s": 57.68400273984298,
      "wall_time_s": 437.66559608,
      "timing": {
        "wall_time_s": 437.66559608,
        "simulation_time_s": 57.68400273984298,
        "physics_steps": 14421,
        "wall_time_start_ns": 1794702709212585721,
        "wall_time_end_ns": 1794703146878201764,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
      "native_instruction": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 156 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-35-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 5,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 3.925,
        "native_frames": 157,
        "published_frames": 157,
        "decode_ok": true,
        "sha256": "70063e7445c8e3924a3bf1df2ce95dd6f21e5701ad696f25c1d8b18f06500dc1",
        "source_sha256": "e17805083e10d1d92e0990562c88f9ab4cf4ea6a17f173b626dff242327582ae",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-36-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/36",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "36",
      "seed": 100000,
      "source_episode_id": "36-place-fan-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 125.83200597669929,
      "wall_time_s": 970.949260869,
      "timing": {
        "wall_time_s": 970.949260869,
        "simulation_time_s": 125.83200597669929,
        "physics_steps": 31458,
        "wall_time_start_ns": 1794703036292128742,
        "wall_time_end_ns": 1794704007241393200,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
      "native_instruction": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-36-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "75eaafdf43860a4ed1e2fa0c18da966948c145f8ae9fc9fa62b32fb4eed6db21",
        "source_sha256": "a85b35a02808b18f8b291397226a2d1d276f1b994979425d07f61b45eca1fd91",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-37-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/37",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "37",
      "seed": 100000,
      "source_episode_id": "37-place-mouse-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 97.66400463879108,
      "wall_time_s": 980.659534012,
      "timing": {
        "wall_time_s": 980.659534012,
        "simulation_time_s": 97.66400463879108,
        "physics_steps": 24416,
        "wall_time_start_ns": 1794703398301182525,
        "wall_time_end_ns": 1794704378960720514,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
      "native_instruction": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-37-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "3480ba663207bd70a98ac8df23a1528d79bf364e52ba27a44b716cc76e845c3f",
        "source_sha256": "870f5935ba02d5a6ce80448dc6b7c302fc4988d44fa9da99302d02d348421692",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-38-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/38",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "38",
      "seed": 100000,
      "source_episode_id": "38-place-object-basket-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 240.05601140204817,
      "wall_time_s": 1900.652805051,
      "timing": {
        "wall_time_s": 1900.652805051,
        "simulation_time_s": 240.05601140204817,
        "physics_steps": 60014,
        "wall_time_start_ns": 1794703849430377858,
        "wall_time_end_ns": 1794705750083186603,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
      "native_instruction": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-38-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "1b6bf3204d439c8f1784f6bdc67fce79f8bf1fd1900f4bd0367c427fad5885d7",
        "source_sha256": "ef3d7272684f85fbc7981e5bb29f6b90b0a28c39fa75282419c122144c7e026c",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-39-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/39",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "39",
      "seed": 100000,
      "source_episode_id": "39-place-object-scale-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 113.24800537899137,
      "wall_time_s": 1043.279987228,
      "timing": {
        "wall_time_s": 1043.279987228,
        "simulation_time_s": 113.24800537899137,
        "physics_steps": 28312,
        "wall_time_start_ns": 1794704390268753777,
        "wall_time_end_ns": 1794705433548745355,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
      "native_instruction": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-39-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "735b84128f551e5c3eb09fd189a72ffe228be5413902aefffc37cea4425e3e97",
        "source_sha256": "fdb6f3025fc388268c4d7c1055c45b0a1328fb0241d758b485c38980cce3c835",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-40-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/40",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "40",
      "seed": 100001,
      "source_episode_id": "40-place-object-stand-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 118
      },
      "steps": 118,
      "simulation_time_s": 36.644001740496606,
      "wall_time_s": 120.820645513,
      "timing": {
        "wall_time_s": 120.820645513,
        "simulation_time_s": 36.644001740496606,
        "physics_steps": 9161,
        "wall_time_start_ns": 1794704310830465343,
        "wall_time_end_ns": 1794704431651120076,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the white stapler on the compact black hollow displaystand with the left arm.",
      "native_instruction": "Place the white stapler on the compact black hollow displaystand with the left arm.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 118 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-40-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 2.975,
        "native_frames": 119,
        "published_frames": 119,
        "decode_ok": true,
        "sha256": "b3fb03ac46f8fcffbcd3897bdc16fa788e9630887d59f0fa1d6eded8c8676df3",
        "source_sha256": "0f2241c892bcf2b25af54a6d31c58fcef2aea4b4409c5f56b25a96c2129559d5",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-41-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/41",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "41",
      "seed": 100000,
      "source_episode_id": "41-place-phone-stand-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 96.15600456716493,
      "wall_time_s": 942.610332014,
      "timing": {
        "wall_time_s": 942.610332014,
        "simulation_time_s": 96.15600456716493,
        "physics_steps": 24039,
        "wall_time_start_ns": 1794704326111896561,
        "wall_time_end_ns": 1794705268722234929,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Set the phone with flat black screen on the angled brown phonestand.",
      "native_instruction": "Set the phone with flat black screen on the angled brown phonestand.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-41-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "e470405872968c5fc7f7b3fbe77f64a16b90a1393c1cb6362892b98d298e453d",
        "source_sha256": "222a82517ff7a64c92a0c1430436d1cf3d68dfe44ae8fb07f80ce4a9f5a3ab2b",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-42-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/42",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "42",
      "seed": 100001,
      "source_episode_id": "42-place-shoe-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 127
      },
      "steps": 127,
      "simulation_time_s": 42.55600202130154,
      "wall_time_s": 210.271377677,
      "timing": {
        "wall_time_s": 210.271377677,
        "simulation_time_s": 42.55600202130154,
        "physics_steps": 10639,
        "wall_time_start_ns": 1794704337772546662,
        "wall_time_end_ns": 1794704548043931440,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the medium adult shoe from the table and set it on the mat",
      "native_instruction": "Pick up the medium adult shoe from the table and set it on the mat",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 127 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-42-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 3.2,
        "native_frames": 128,
        "published_frames": 128,
        "decode_ok": true,
        "sha256": "621656656c98df9dbecc22cc8c3e5489fc217c521ffe62f5bf9a9e2d708815a7",
        "source_sha256": "878b97025124d90256798944187fbbaac9861e7098688fce3ce5071d74fb1ce8",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-43-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/43",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "43",
      "seed": 100000,
      "source_episode_id": "43-press-stapler-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 344
      },
      "steps": 344,
      "simulation_time_s": 63.01600299309939,
      "wall_time_s": 909.878646164,
      "timing": {
        "wall_time_s": 909.878646164,
        "simulation_time_s": 63.01600299309939,
        "physics_steps": 15754,
        "wall_time_start_ns": 1794704398048935888,
        "wall_time_end_ns": 1794705307927587347,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Press the plastic stapler firmly using your arm.",
      "native_instruction": "Press the plastic stapler firmly using your arm.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 344 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-43-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 11,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 8.625,
        "native_frames": 345,
        "published_frames": 345,
        "decode_ok": true,
        "sha256": "da19e2e52ac1fb0758e18eb89f47cedc7d2ea06e70e4fefa3d12587c43135bd7",
        "source_sha256": "cd73d71c92e9a110f33613dbb830a2289a66b114b7679233c753520aed878ee3",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-44-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/44",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "44",
      "seed": 100000,
      "source_episode_id": "44-rotate-qrcode-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 117.22800556803122,
      "wall_time_s": 1055.97267256,
      "timing": {
        "wall_time_s": 1055.97267256,
        "simulation_time_s": 117.22800556803122,
        "physics_steps": 29307,
        "wall_time_start_ns": 1794704469978016564,
        "wall_time_end_ns": 1794705525950693561,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
      "native_instruction": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-44-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "49f943ff5d6b30075b60b3a726f1b6a5c7791cc5fb691e5b5f298b2eb539c12c",
        "source_sha256": "3c0b6bbda61f387bfb4d60eef0cabef725315a8676e2a44f281c424dbd12e995",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-45-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/45",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "45",
      "seed": 100001,
      "source_episode_id": "45-shake-bottle-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 61
      },
      "steps": 61,
      "simulation_time_s": 21.11600100295618,
      "wall_time_s": 305.879945322,
      "timing": {
        "wall_time_s": 305.879945322,
        "simulation_time_s": 21.11600100295618,
        "physics_steps": 5279,
        "wall_time_start_ns": 1794704600399779284,
        "wall_time_end_ns": 1794704906279730523,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the smooth bottle with blue text and shake it.",
      "native_instruction": "Pick up the smooth bottle with blue text and shake it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 61 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-45-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 2,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.55,
        "native_frames": 62,
        "published_frames": 62,
        "decode_ok": true,
        "sha256": "3e51a41dab7d43d8fb830aefc26ed24979220c78c50328959500aa83a3e6e148",
        "source_sha256": "9befefa273263d76ffbdda2756b0c59220687b672db47e6e2cc9dcfbf773839a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-46-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/46",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "46",
      "seed": 100001,
      "source_episode_id": "46-shake-bottle-horizontally-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 69
      },
      "steps": 69,
      "simulation_time_s": 24.912001183256507,
      "wall_time_s": 262.015192753,
      "timing": {
        "wall_time_s": 262.015192753,
        "simulation_time_s": 24.912001183256507,
        "physics_steps": 6228,
        "wall_time_start_ns": 1794704568774920315,
        "wall_time_end_ns": 1794704830790123807,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
      "native_instruction": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 69 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-46-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.75,
        "native_frames": 70,
        "published_frames": 70,
        "decode_ok": true,
        "sha256": "1832b83db1a1d21d5143b0af1f8bbbcb935c9f5b38a184859df6f5b12a3ad83d",
        "source_sha256": "02607da443ae61a07697f32b624b8376dbd6b0ccebcfde4e22f91cf9a2c628c1",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-47-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/47",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "47",
      "seed": 100000,
      "source_episode_id": "47-stack-blocks-two-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 274
      },
      "steps": 274,
      "simulation_time_s": 84.76800402626395,
      "wall_time_s": 754.705034343,
      "timing": {
        "wall_time_s": 754.705034343,
        "simulation_time_s": 84.76800402626395,
        "physics_steps": 21192,
        "wall_time_start_ns": 1794704666683885104,
        "wall_time_end_ns": 1794705421388924720,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Move red block to the center, then put green block on top.",
      "native_instruction": "Move red block to the center, then put green block on top.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 274 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-47-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 9,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 6.875,
        "native_frames": 275,
        "published_frames": 275,
        "decode_ok": true,
        "sha256": "54a6fed8987d2d1f752860823764c9234325db5048c9adb68ff77433e6d74ec5",
        "source_sha256": "cd216d69e6e1fe6e826a414c9d4d7782bea8a1517b16ce00114cd7cc175e83ad",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-48-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/48",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "48",
      "seed": 100000,
      "source_episode_id": "48-stack-bowls-two-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 315
      },
      "steps": 315,
      "simulation_time_s": 96.74800459528342,
      "wall_time_s": 845.522166104,
      "timing": {
        "wall_time_s": 845.522166104,
        "simulation_time_s": 96.74800459528342,
        "physics_steps": 24187,
        "wall_time_start_ns": 1794704735001790331,
        "wall_time_end_ns": 1794705580523963367,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
      "native_instruction": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 315 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-48-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 10,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 7.9,
        "native_frames": 316,
        "published_frames": 316,
        "decode_ok": true,
        "sha256": "6fed07971af60cf01a74a6484b3800ff37a0ad91a88f0239a81b3c4afff46422",
        "source_sha256": "9b17ef0366d0e5877321afd0495c135c3a1c66807ea4a00fdd637fb0e4e96e70",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-49-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/49",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "49",
      "seed": 100003,
      "source_episode_id": "49-stamp-seal-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 126.44800600595772,
      "wall_time_s": 1136.678702267,
      "timing": {
        "wall_time_s": 1136.678702267,
        "simulation_time_s": 126.44800600595772,
        "physics_steps": 31612,
        "wall_time_start_ns": 1794704814167495865,
        "wall_time_end_ns": 1794705950846201453,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the block-shaped seal with angled edges, press onto Blue",
      "native_instruction": "Grab the block-shaped seal with angled edges, press onto Blue",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-49-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100003,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "cff5ed17309639b968feec73022103e3e2b00d846539ef218a2b81fcc1a051af",
        "source_sha256": "44eeafef70402e6900a316589f33a38dcc2703393e220b5a6d72bf14b03bddf3",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-jy-reference_pose-50-20261113",
      "variant": "jy-mode2",
      "task_key": "task03/50",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "50",
      "seed": 100000,
      "source_episode_id": "50-turn-switch-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 94.74400450009853,
      "wall_time_s": 814.158012429,
      "timing": {
        "wall_time_s": 814.158012429,
        "simulation_time_s": 94.74400450009853,
        "physics_steps": 23686,
        "wall_time_start_ns": 1794704918256217895,
        "wall_time_end_ns": 1794705732414233379,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-jy",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Press the flat square gray switch using the left arm",
      "native_instruction": "Press the flat square gray switch using the left arm",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-jy-reference_pose-50-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "3baf49a272f2d5814444c7e07f2a412ab1009eedad75f11a5510dc076935cfc7",
        "source_sha256": "6b671981a9c5a4d0eb68a7141ad4c7c9c2d8ae1a630b4f30c859565556c8df2e",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-01-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/01",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "01",
      "seed": 100001,
      "source_episode_id": "01-handover-block-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 800
      },
      "steps": 800,
      "simulation_time_s": 505.14402399305254,
      "wall_time_s": 2154.358827509,
      "timing": {
        "wall_time_s": 2154.358827509,
        "simulation_time_s": 505.14402399305254,
        "physics_steps": 126286,
        "wall_time_start_ns": 1794687686147687356,
        "wall_time_end_ns": 1794689840506517955,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
      "native_instruction": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 800 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-01-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 25,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 20.0,
        "native_frames": 800,
        "published_frames": 800,
        "decode_ok": true,
        "sha256": "1d7e9b37b005c07db719ed6cd94ba3776f9a7288daaa7e0338ce7eee90c0055f",
        "source_sha256": "6ec772637bf725e8a96741c16b82c5ec4175e25ebf0af7cbf95fd265d833fd8d",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-02-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/02",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "02",
      "seed": 100001,
      "source_episode_id": "02-place-dual-shoes-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 600
      },
      "steps": 600,
      "simulation_time_s": 238.10001130914316,
      "wall_time_s": 1952.106195306,
      "timing": {
        "wall_time_s": 1952.106195306,
        "simulation_time_s": 238.10001130914316,
        "physics_steps": 59525,
        "wall_time_start_ns": 1794687783783308697,
        "wall_time_end_ns": 1794689735889514403,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
      "native_instruction": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 600 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-02-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 15.0,
        "native_frames": 600,
        "published_frames": 600,
        "decode_ok": true,
        "sha256": "bc6b976078b60e2e5dd689a805d7cc48fb9a571428a905991d3fd42eac239591",
        "source_sha256": "fdc1864794fa1afd0ffdc0dc9dd3ad1cccee6059ec6296c8d0fa46c056f223ad",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-03-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/03",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "03",
      "seed": 100000,
      "source_episode_id": "03-put-bottles-dustbin-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1700
      },
      "steps": 1700,
      "simulation_time_s": 576.6240273881704,
      "wall_time_s": 4474.39121597,
      "timing": {
        "wall_time_s": 4474.39121597,
        "simulation_time_s": 576.6240273881704,
        "physics_steps": 144156,
        "wall_time_start_ns": 1794687809893191415,
        "wall_time_end_ns": 1794692284284410549,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
      "native_instruction": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-03-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1700,
        "max_control_steps": 1700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 54,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 42.5,
        "native_frames": 1700,
        "published_frames": 1700,
        "decode_ok": true,
        "sha256": "293c3f20a42e0d3e100c5f381cd93387d7dbad9c901db5ce4e88b11a3a2113a9",
        "source_sha256": "186cd09d7d5da5ee16bcbd3f6cc73484c5b53c59ab99158ea3a56f6cec35e6a3",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184,
          1216,
          1248,
          1280,
          1312,
          1344,
          1376,
          1408,
          1440,
          1472,
          1504,
          1536,
          1568,
          1600,
          1632,
          1664,
          1696
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-04-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/04",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "04",
      "seed": 100001,
      "source_episode_id": "04-scan-object-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 183.5840087197721,
      "wall_time_s": 1804.437891399,
      "timing": {
        "wall_time_s": 1804.437891399,
        "simulation_time_s": 183.5840087197721,
        "physics_steps": 45896,
        "wall_time_start_ns": 1794687844292688638,
        "wall_time_end_ns": 1794689648730584717,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
      "native_instruction": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-04-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "71689248479bea9428f90ec11f7b85c32a4909e26c15fe2398717bb3c1a68d9e",
        "source_sha256": "0471fa0a34defa267147cb77a78b50426c6bfb4d2c7369ab6e4ecebccf94e773",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-05-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/05",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "05",
      "seed": 100000,
      "source_episode_id": "05-stack-bowls-three-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 438.36002082098275,
      "wall_time_s": 3954.16485034,
      "timing": {
        "wall_time_s": 3954.16485034,
        "simulation_time_s": 438.36002082098275,
        "physics_steps": 109590,
        "wall_time_start_ns": 1794687929370189003,
        "wall_time_end_ns": 1794691883535044601,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
      "native_instruction": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-05-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "a081a612b93871ca67de31068973a57d4055afc974ab565d9820476ff7476a12",
        "source_sha256": "abca82038884c682f5f94d9f02ba823963e9d10b89733ae0f89d2c26cd38bfdc",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-06-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/06",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "06",
      "seed": 100000,
      "source_episode_id": "06-stack-blocks-three-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 650.1080308784731,
      "wall_time_s": 3929.272123191,
      "timing": {
        "wall_time_s": 3929.272123191,
        "simulation_time_s": 650.1080308784731,
        "physics_steps": 162527,
        "wall_time_start_ns": 1794688035144312605,
        "wall_time_end_ns": 1794691964416440149,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
      "native_instruction": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-06-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "a9eb7018e064a399b1749506bd7217487905a5fdab07d08abaeaa8f49573dbd0",
        "source_sha256": "83ee6afe8b2a9ea47a07b37043020d9914903104274a6cb90a19a39f88245066",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-07-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/07",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "07",
      "seed": 100000,
      "source_episode_id": "07-hanging-mug-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 900
      },
      "steps": 900,
      "simulation_time_s": 430.5440204497427,
      "wall_time_s": 2875.051569754,
      "timing": {
        "wall_time_s": 2875.051569754,
        "simulation_time_s": 430.5440204497427,
        "physics_steps": 107636,
        "wall_time_start_ns": 1794688016799513024,
        "wall_time_end_ns": 1794690891851088700,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
      "native_instruction": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 900 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-07-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 29,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 22.5,
        "native_frames": 900,
        "published_frames": 900,
        "decode_ok": true,
        "sha256": "9192a6099071388b3dadfa9a2dfd2e76c7eb3545032d19da4b1cc1440e724386",
        "source_sha256": "d13ea9fc38eda8a6399be9894417c93173ba586b9080beab95c8f6b43e0c45e5",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-08-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/08",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "08",
      "seed": 100001,
      "source_episode_id": "08-put-object-cabinet-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 250.9120119176805,
      "wall_time_s": 2654.006547198,
      "timing": {
        "wall_time_s": 2654.006547198,
        "simulation_time_s": 250.9120119176805,
        "physics_steps": 62728,
        "wall_time_start_ns": 1794688403720493310,
        "wall_time_end_ns": 1794691057727045345,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
      "native_instruction": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-08-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "e036a3c5fda4536f34bb74b7934887446fb6fd5cedc8ec93a52e4dcc8bd60b15",
        "source_sha256": "797b6696d7832227659ee23b2200c18e6eb1867e2bce2fcf2f8a16eeca590b2b",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-09-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/09",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "09",
      "seed": 100000,
      "source_episode_id": "09-blocks-ranking-size-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 479.29202276514843,
      "wall_time_s": 2365.485569981,
      "timing": {
        "wall_time_s": 2365.485569981,
        "simulation_time_s": 479.29202276514843,
        "physics_steps": 119823,
        "wall_time_start_ns": 1794689992188525787,
        "wall_time_end_ns": 1794692357674098249,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place large block, medium block, and small block in order at the table's center.",
      "native_instruction": "Place large block, medium block, and small block in order at the table's center.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-09-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "ccc16d3e5b013b1773962f7f7932727047da3c411d1ac26a7f3f2a20d3cf1b60",
        "source_sha256": "1ecdb1c8626f5f11f62ba9c4cb863b5152dd016c5f151b77711c4be5321d26cc",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-10-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/10",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "10",
      "seed": 100000,
      "source_episode_id": "10-lift-pot-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 142.32400676002726,
      "wall_time_s": 1495.892642653,
      "timing": {
        "wall_time_s": 1495.892642653,
        "simulation_time_s": 142.32400676002726,
        "physics_steps": 35581,
        "wall_time_start_ns": 1794690046656232128,
        "wall_time_end_ns": 1794691542548877700,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the dark gray cooking kitchenpot using both arms",
      "native_instruction": "Lift the dark gray cooking kitchenpot using both arms",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-10-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "5ab2532f6b18c21cf91655a4b3b9683956ed565bc24daf12efd64159c61d9779",
        "source_sha256": "ca605636348a3bff48508453bbe2da6071d57d71c0b41442eccb97ed08d23c23",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-11-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/11",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "11",
      "seed": 100000,
      "source_episode_id": "11-adjust-bottle-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 196.32000932469964,
      "wall_time_s": 1473.74284737,
      "timing": {
        "wall_time_s": 1473.74284737,
        "simulation_time_s": 196.32000932469964,
        "physics_steps": 49080,
        "wall_time_start_ns": 1794690162664546357,
        "wall_time_end_ns": 1794691636407398510,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the brown bottle upright",
      "native_instruction": "Use the left arm to grab the brown bottle upright",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-11-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "4f25b9f9f401ed3fe1090eed70f06564d1cc8c4a35ffd1a37ba5a8f0a43be016",
        "source_sha256": "bec83309f58afb60104be90611e794174c8aeebfcda63e7ca2e8bc043b6db542",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-12-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/12",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "12",
      "seed": 100000,
      "source_episode_id": "12-beat-block-hammer-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 183.79200872965157,
      "wall_time_s": 737.22135925,
      "timing": {
        "wall_time_s": 737.22135925,
        "simulation_time_s": 183.79200872965157,
        "physics_steps": 45948,
        "wall_time_start_ns": 1794690270866920514,
        "wall_time_end_ns": 1794691008088284125,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the hammer with black handle then beat block",
      "native_instruction": "Use the left arm to grab the hammer with black handle then beat block",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-12-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "d98322d9e260c15ce5c08cabec4b07bee4f4de2f01a44e8bcf2a50689fdbdf96",
        "source_sha256": "01da26b9b720df8967c2ce518cd654366c4616bc8461d4be84a4fb4ba0bddb43",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-13-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/13",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "13",
      "seed": 100000,
      "source_episode_id": "13-blocks-ranking-rgb-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 453.5360215418041,
      "wall_time_s": 1686.808889955,
      "timing": {
        "wall_time_s": 1686.808889955,
        "simulation_time_s": 453.5360215418041,
        "physics_steps": 113384,
        "wall_time_start_ns": 1794690295070717858,
        "wall_time_end_ns": 1794691981879610682,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
      "native_instruction": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-13-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "ec88356f6fbc4505d41f51d321d4c9de7823d082ce126d28ca49602c0baae39c",
        "source_sha256": "5f84300ac1ef3833b7ac46147d792187988f5416341530fe02e6a28a7c21ae1b",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-14-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/14",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "14",
      "seed": 100000,
      "source_episode_id": "14-click-alarmclock-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 100.58800477767363,
      "wall_time_s": 440.672902136,
      "timing": {
        "wall_time_s": 440.672902136,
        "simulation_time_s": 100.58800477767363,
        "physics_steps": 25147,
        "wall_time_start_ns": 1794690338091712128,
        "wall_time_end_ns": 1794690778764617952,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Tap the center top button of the alarm-clock with beige screen",
      "native_instruction": "Tap the center top button of the alarm-clock with beige screen",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-14-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "00c96474d8646ee616b9bc54b710ac5e3d0c26aeff7b47b4b3074ef7ab82a9ee",
        "source_sha256": "3b6837631d1936cb0ed25519fdae7200c505dbc2fda007b18318c9b684400224",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-15-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/15",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "15",
      "seed": 100000,
      "source_episode_id": "15-click-bell-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 81
      },
      "steps": 81,
      "simulation_time_s": 43.836002082098275,
      "wall_time_s": 278.458462339,
      "timing": {
        "wall_time_s": 278.458462339,
        "simulation_time_s": 43.836002082098275,
        "physics_steps": 10959,
        "wall_time_start_ns": 1794690486611371958,
        "wall_time_end_ns": 1794690765069839382,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to press the medium-sized bell's top center.",
      "native_instruction": "Use the left arm to press the medium-sized bell's top center.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 81 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-15-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 2.05,
        "native_frames": 82,
        "published_frames": 82,
        "decode_ok": true,
        "sha256": "3c4d5d967ae31a9a0960afe45a641e74b03e5316b0fc50b6389b6014ad0bccff",
        "source_sha256": "8674c0c997a93bab08d6c5286e9c081bd272c6a45646a5b098cfe74846c3c2f0",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-16-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/16",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "16",
      "seed": 100006,
      "source_episode_id": "16-dump-bin-bigbin-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 600
      },
      "steps": 600,
      "simulation_time_s": 330.7280157087371,
      "wall_time_s": 1030.478488169,
      "timing": {
        "wall_time_s": 1030.478488169,
        "simulation_time_s": 330.7280157087371,
        "physics_steps": 82682,
        "wall_time_start_ns": 1794690582584197572,
        "wall_time_end_ns": 1794691613062690076,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
      "native_instruction": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 600 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-16-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100006,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 15.0,
        "native_frames": 600,
        "published_frames": 600,
        "decode_ok": true,
        "sha256": "15e8843d226f053fc1b1a27987aa9cc2a032693fbb2d8454fe3743f251d75cf7",
        "source_sha256": "30ccceaa4ec166025685e0caba76b37eb41a7ae4998722245116d7d085c6e9e4",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-17-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/17",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "17",
      "seed": 100000,
      "source_episode_id": "17-grab-roller-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 153.83600730681792,
      "wall_time_s": 799.874760618,
      "timing": {
        "wall_time_s": 799.874760618,
        "simulation_time_s": 153.83600730681792,
        "physics_steps": 38459,
        "wall_time_start_ns": 1794698088137711337,
        "wall_time_end_ns": 1794698888012475536,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use arms to grab the smooth wooden roller.",
      "native_instruction": "Use arms to grab the smooth wooden roller.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-17-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "8530efd85060a7eb957d27e18ff6700a7e0e61a028b0f36f03884efa04a61ea5",
        "source_sha256": "8e26b892680b23c116eb080d44767e1becacdb400cb51cad2b7205d09cb41f78",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-18-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/18",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "18",
      "seed": 100000,
      "source_episode_id": "18-handover-mic-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 600
      },
      "steps": 600,
      "simulation_time_s": 264.2200125497766,
      "wall_time_s": 1346.807587724,
      "timing": {
        "wall_time_s": 1346.807587724,
        "simulation_time_s": 264.2200125497766,
        "physics_steps": 66055,
        "wall_time_start_ns": 1794698107100949624,
        "wall_time_end_ns": 1794699453908547153,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use one arm to hold the microphone used for audio input and hand it over.",
      "native_instruction": "Use one arm to hold the microphone used for audio input and hand it over.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 600 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-18-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 15.0,
        "native_frames": 600,
        "published_frames": 600,
        "decode_ok": true,
        "sha256": "f86c025423b658792112f09cad83d6025d9a14e72f72e93d8282d5fb8b4678c6",
        "source_sha256": "4871d5a2c96d3f8c3798d1a11c899278eec300fd2962b82bacef30ba0dbd3c42",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-19-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/19",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "19",
      "seed": 100000,
      "source_episode_id": "19-move-can-pot-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 207.51200985629112,
      "wall_time_s": 1046.599455875,
      "timing": {
        "wall_time_s": 1046.599455875,
        "simulation_time_s": 207.51200985629112,
        "physics_steps": 51878,
        "wall_time_start_ns": 1794698159897396052,
        "wall_time_end_ns": 1794699206496855812,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
      "native_instruction": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-19-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "8fa4ebf78fd5704867870d4aca3990c7b1f56e5ad8cf4382271172097a48312b",
        "source_sha256": "44ec2f22a63470a1f4a0dc72f7d51acc67302475c88e97828cb81f65f6127b19",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-20-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/20",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "20",
      "seed": 100001,
      "source_episode_id": "20-move-pillbottle-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 223.96401063771918,
      "wall_time_s": 1138.963365007,
      "timing": {
        "wall_time_s": 1138.963365007,
        "simulation_time_s": 223.96401063771918,
        "physics_steps": 55991,
        "wall_time_start_ns": 1794698267512348211,
        "wall_time_end_ns": 1794699406475717447,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the white cylindrical bottle and set it on the pad",
      "native_instruction": "Grab the white cylindrical bottle and set it on the pad",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-20-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "7aa0cfac0db8a3cd983c4103e93e3394e6578b9f5fe7121584bd9e1d04e40a3e",
        "source_sha256": "7856a7322a7f020ed8f3b854f87148991a3235848c28fe2f2eb199ffb7cfe6f8",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-21-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/21",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "21",
      "seed": 100000,
      "source_episode_id": "21-move-playingcard-away-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 150.90800716774538,
      "wall_time_s": 1270.459331324,
      "timing": {
        "wall_time_s": 1270.459331324,
        "simulation_time_s": 150.90800716774538,
        "physics_steps": 37727,
        "wall_time_start_ns": 1794698302818730757,
        "wall_time_end_ns": 1794699573278066544,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the right arm to grab the hard black box for card deck and relocate it.",
      "native_instruction": "Use the right arm to grab the hard black box for card deck and relocate it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-21-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "4071fce75b21c4091864393e19adaedfa2158d7f8309847428c05909561e090d",
        "source_sha256": "26a789dde9d46e4d2376397a6d7a9c81fc8a98d6dfee5e0c1642dedd5bcbc665",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-22-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/22",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "22",
      "seed": 100000,
      "source_episode_id": "22-move-stapler-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 192.98800916643813,
      "wall_time_s": 1039.309138855,
      "timing": {
        "wall_time_s": 1039.309138855,
        "simulation_time_s": 192.98800916643813,
        "physics_steps": 48247,
        "wall_time_start_ns": 1794698549702024541,
        "wall_time_end_ns": 1794699589011167466,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to move the plastic stapler to Yellow mat",
      "native_instruction": "Use the left arm to move the plastic stapler to Yellow mat",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-22-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "60c99166547c2a1ad4c48afd97d429b1264c332de01c4b0b7d92996dc74e81de",
        "source_sha256": "b99b6a31d8ff06071721db4554b6c07102b8b75edb10bd2b05af72dd74e0f9fd",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-23-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/23",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "23",
      "seed": 100000,
      "source_episode_id": "23-open-laptop-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 245.8520116773434,
      "wall_time_s": 2215.96059262,
      "timing": {
        "wall_time_s": 2215.96059262,
        "simulation_time_s": 245.8520116773434,
        "physics_steps": 61463,
        "wall_time_start_ns": 1794699241876652279,
        "wall_time_end_ns": 1794701457837249453,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Open the flat orange laptop with one motion.",
      "native_instruction": "Open the flat orange laptop with one motion.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-23-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "11c40bca398fcb710e20f301fbdd7b32fdef182719e7935f7e0ca57821cb70ce",
        "source_sha256": "ad5e9c130a0289c2ec2f1261d27b97bafc62b3e087836bf288a80a0ac81a938a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-24-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/24",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "24",
      "seed": 100000,
      "source_episode_id": "24-open-microwave-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1500
      },
      "steps": 1500,
      "simulation_time_s": 670.4840318462811,
      "wall_time_s": 3636.393544577,
      "timing": {
        "wall_time_s": 3636.393544577,
        "simulation_time_s": 670.4840318462811,
        "physics_steps": 167621,
        "wall_time_start_ns": 1794699479016664258,
        "wall_time_end_ns": 1794703115410212461,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
      "native_instruction": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-24-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1500,
        "max_control_steps": 1500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 47,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 37.5,
        "native_frames": 1500,
        "published_frames": 1500,
        "decode_ok": true,
        "sha256": "5d72c4168ba1d131c1488aa4e0f5101814d4bb4356a5a56a67eb0879d14b5414",
        "source_sha256": "ea0276dea16a738701d1788df4e00d5d4c2d1bf24c7296e0a535e556a166434f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184,
          1216,
          1248,
          1280,
          1312,
          1344,
          1376,
          1408,
          1440,
          1472
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-25-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/25",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "25",
      "seed": 100005,
      "source_episode_id": "25-pick-diverse-bottles-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 172.46000819141045,
      "wall_time_s": 1495.164826343,
      "timing": {
        "wall_time_s": 1495.164826343,
        "simulation_time_s": 172.46000819141045,
        "physics_steps": 43115,
        "wall_time_start_ns": 1794699767583859604,
        "wall_time_end_ns": 1794701262748689807,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
      "native_instruction": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-25-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100005,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "907d97b29f1df2bcea3b3c38a7d0f5ca2da3bce3f50187920121b9eee66135f2",
        "source_sha256": "7571134299e75678dfec43a9bfa6848a404cd3c9deeac43efa773ef3e90ff63a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-26-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/26",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "26",
      "seed": 100000,
      "source_episode_id": "26-pick-dual-bottles-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 172.12000817526132,
      "wall_time_s": 1060.668086445,
      "timing": {
        "wall_time_s": 1060.668086445,
        "simulation_time_s": 172.12000817526132,
        "physics_steps": 43030,
        "wall_time_start_ns": 1794699619363550084,
        "wall_time_end_ns": 1794700680031640916,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
      "native_instruction": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-26-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "3e053453a0268f07498163ca05272af446d120666d04ffb6b12ee1618b27ec87",
        "source_sha256": "7c70bd12611de7fedb8ffc647d7011412387f917945a0c348ce6313004ebf481",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-27-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/27",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "27",
      "seed": 100000,
      "source_episode_id": "27-place-a2b-left-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 182.7240086789243,
      "wall_time_s": 1106.236252502,
      "timing": {
        "wall_time_s": 1106.236252502,
        "simulation_time_s": 182.7240086789243,
        "physics_steps": 45681,
        "wall_time_start_ns": 1794699774305299966,
        "wall_time_end_ns": 1794700880541557012,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
      "native_instruction": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-27-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "02c78d04edb71f3fcbc2c7ee2a7db87451df256ba07fdbc0c8de5446347f9360",
        "source_sha256": "6e9b6f6ed74a32510f50a88ece08035d1f838f35615e28330dbe8f7b63c8d909",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-28-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/28",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "28",
      "seed": 100000,
      "source_episode_id": "28-place-a2b-right-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 189.60400900570676,
      "wall_time_s": 1134.070099495,
      "timing": {
        "wall_time_s": 1134.070099495,
        "simulation_time_s": 189.60400900570676,
        "physics_steps": 47401,
        "wall_time_start_ns": 1794700122927235712,
        "wall_time_end_ns": 1794701256997339730,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
      "native_instruction": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-28-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "076adc7d983031627854cfdca63b168b73df482e7beb0612eacdc8c84b9550a2",
        "source_sha256": "68c0ea994a091ed7e8f91f5989548a9b9bca118308db4c9ed37ab798c2f75dc0",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-29-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/29",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "29",
      "seed": 100000,
      "source_episode_id": "29-place-bread-basket-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 291.22401383239776,
      "wall_time_s": 2169.172629698,
      "timing": {
        "wall_time_s": 2169.172629698,
        "simulation_time_s": 291.22401383239776,
        "physics_steps": 72806,
        "wall_time_start_ns": 1794701241621512051,
        "wall_time_end_ns": 1794703410794148716,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
      "native_instruction": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-29-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "a34df45567877a23bd569d3c0ba860cdba3ee69936610d16c93308c227ed55db",
        "source_sha256": "2726636877522de555c204dd08841b854c8624285e03aff4297168f255a7002a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-30-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/30",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "30",
      "seed": 100001,
      "source_episode_id": "30-place-bread-skillet-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 199.07600945560262,
      "wall_time_s": 1396.651262173,
      "timing": {
        "wall_time_s": 1396.651262173,
        "simulation_time_s": 199.07600945560262,
        "physics_steps": 49769,
        "wall_time_start_ns": 1794701279041710155,
        "wall_time_end_ns": 1794702675692976982,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
      "native_instruction": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-30-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "8272cdbadcfeae08950831150ccc4943439ca8284d79c32b03e3020dca6b7cc9",
        "source_sha256": "e184d122664763686b74b0eb77f9947c3da8e4be3693ef63e412505c794c9550",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-31-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/31",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "31",
      "seed": 100000,
      "source_episode_id": "31-place-burger-fries-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 218.76401039073244,
      "wall_time_s": 1661.313860337,
      "timing": {
        "wall_time_s": 1661.313860337,
        "simulation_time_s": 218.76401039073244,
        "physics_steps": 54691,
        "wall_time_start_ns": 1794701708098039790,
        "wall_time_end_ns": 1794703369411905255,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
      "native_instruction": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-31-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "a8a9089de66a6dfbd86c530ae587271bff0daa85d181dda27e508b5a2ceb3723",
        "source_sha256": "42f2d46c9859d59cd18696d922e8e22ea06ab785d81a82c37bd7abd2cb10514f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-32-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/32",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "32",
      "seed": 100000,
      "source_episode_id": "32-place-can-basket-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 359.7520170873031,
      "wall_time_s": 1763.869268496,
      "timing": {
        "wall_time_s": 1763.869268496,
        "simulation_time_s": 359.7520170873031,
        "physics_steps": 89938,
        "wall_time_start_ns": 1794701635606886719,
        "wall_time_end_ns": 1794703399476161316,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "native_instruction": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-32-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "0c4e71d88877e6a50e7677eb9a46abc9cf0251847d71fae7be76e0994b6d56fd",
        "source_sha256": "895ade0883c7863747c28a82d82f8c5a584eb5dff12e9e292bea71ba299d3fac",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-33-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/33",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "33",
      "seed": 100000,
      "source_episode_id": "33-place-cans-plasticbox-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 800
      },
      "steps": 800,
      "simulation_time_s": 395.20801877137274,
      "wall_time_s": 2540.647608508,
      "timing": {
        "wall_time_s": 2540.647608508,
        "simulation_time_s": 395.20801877137274,
        "physics_steps": 98802,
        "wall_time_start_ns": 1794702023984412952,
        "wall_time_end_ns": 1794704564632026402,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
      "native_instruction": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 800 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-33-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 25,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 20.0,
        "native_frames": 800,
        "published_frames": 800,
        "decode_ok": true,
        "sha256": "9be27ba248200c9a23ce093721c92350def36ccda753e446ec6968643cde9d8a",
        "source_sha256": "ea7f7805e680ff64592fa3faad28c94233cb8f6727edc7ff63976b87086d1247",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-34-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/34",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "34",
      "seed": 100000,
      "source_episode_id": "34-place-container-plate-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 207.15200983919203,
      "wall_time_s": 975.338185939,
      "timing": {
        "wall_time_s": 975.338185939,
        "simulation_time_s": 207.15200983919203,
        "physics_steps": 51788,
        "wall_time_start_ns": 1794703068853904881,
        "wall_time_end_ns": 1794704044192096065,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the curved bowl with smooth texture on the gray plate for dining.",
      "native_instruction": "Place the curved bowl with smooth texture on the gray plate for dining.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-34-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "015d3a5d701d578757316c053e0e151bb8da305f05fe3682967acae9e68a5d2f",
        "source_sha256": "ebf84129405f3f78da46d3db8565110b1b4fe2e433c6c09882f83e2876e44fb0",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-35-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/35",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "35",
      "seed": 100000,
      "source_episode_id": "35-place-empty-cup-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 255.49601213540882,
      "wall_time_s": 1563.042451003,
      "timing": {
        "wall_time_s": 1563.042451003,
        "simulation_time_s": 255.49601213540882,
        "physics_steps": 63874,
        "wall_time_start_ns": 1794703398415634784,
        "wall_time_end_ns": 1794704961458090037,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
      "native_instruction": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-35-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "7f32431f2b630437b8ec095acdd7306c6473c5cfd81e35ff2b1d84506bde4f12",
        "source_sha256": "6dbae8bb014197310669f52b3cccbf4b5505fff0b7bc7fee415900845fb3defd",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-36-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/36",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "36",
      "seed": 100000,
      "source_episode_id": "36-place-fan-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 189.06800898024812,
      "wall_time_s": 1023.434691732,
      "timing": {
        "wall_time_s": 1023.434691732,
        "simulation_time_s": 189.06800898024812,
        "physics_steps": 47267,
        "wall_time_start_ns": 1794703590062356531,
        "wall_time_end_ns": 1794704613497053184,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
      "native_instruction": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-36-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "092cb29e4b567dabca3a8ca598be8ad924f7219e9f5f9f5d1bfb1270b42f557e",
        "source_sha256": "e25b2b701af5b2b90d3a327e3a15246031976142527a63710af680e420a5a53f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-37-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/37",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "37",
      "seed": 100000,
      "source_episode_id": "37-place-mouse-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 210.36000999156386,
      "wall_time_s": 1310.031382665,
      "timing": {
        "wall_time_s": 1310.031382665,
        "simulation_time_s": 210.36000999156386,
        "physics_steps": 52590,
        "wall_time_start_ns": 1794703711951648194,
        "wall_time_end_ns": 1794705021983035402,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
      "native_instruction": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-37-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "8a8b048040aa45a3e17ad74605d81a0c4d3c4ec766e611202dd5b0da6355ebc6",
        "source_sha256": "32e35e71d287d0392c7689f21ec0354ed43e8ea5d1c5ccd9e416fc2161bf5828",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-38-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/38",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "38",
      "seed": 100000,
      "source_episode_id": "38-place-object-basket-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 330.6400157045573,
      "wall_time_s": 1772.865329396,
      "timing": {
        "wall_time_s": 1772.865329396,
        "simulation_time_s": 330.6400157045573,
        "physics_steps": 82660,
        "wall_time_start_ns": 1794703712968853351,
        "wall_time_end_ns": 1794705485834187456,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
      "native_instruction": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-38-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "45174b131adf91509d851843cd9a7c7aac72a0e5e97733abee25912ba126a882",
        "source_sha256": "b51f6381f4fdaa8d0a6b85d83c26af9a47e1f9fb7bd862a72dd246ff94f04e01",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-39-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/39",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "39",
      "seed": 100000,
      "source_episode_id": "39-place-object-scale-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 156.32400742499158,
      "wall_time_s": 1037.635689672,
      "timing": {
        "wall_time_s": 1037.635689672,
        "simulation_time_s": 156.32400742499158,
        "physics_steps": 39081,
        "wall_time_start_ns": 1794704325139336040,
        "wall_time_end_ns": 1794705362775030211,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
      "native_instruction": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-39-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "9dadfcae3c5b953a11a7bde2757d897baae7f5134b52391ca21377bf11518549",
        "source_sha256": "31aeca1167ce23825d3dd8ff83730f7c84a8951bdbb2c205eb5227e1062e2ae5",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-40-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/40",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "40",
      "seed": 100001,
      "source_episode_id": "40-place-object-stand-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 157.91600750060752,
      "wall_time_s": 775.178439025,
      "timing": {
        "wall_time_s": 775.178439025,
        "simulation_time_s": 157.91600750060752,
        "physics_steps": 39479,
        "wall_time_start_ns": 1794704320150424260,
        "wall_time_end_ns": 1794705095328867752,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the white stapler on the compact black hollow displaystand with the left arm.",
      "native_instruction": "Place the white stapler on the compact black hollow displaystand with the left arm.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-40-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "c3b49784ece3f2e983dee181b09705c90fe97975f5e0b4b4491c8d0dd14b4b4d",
        "source_sha256": "c1f24d16b4813beb5c2ba9e97ad344e968b3ab0287b004929c3a0b03fec63225",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-41-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/41",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "41",
      "seed": 100000,
      "source_episode_id": "41-place-phone-stand-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 170.47200809698552,
      "wall_time_s": 1003.91589784,
      "timing": {
        "wall_time_s": 1003.91589784,
        "simulation_time_s": 170.47200809698552,
        "physics_steps": 42618,
        "wall_time_start_ns": 1794704355084702255,
        "wall_time_end_ns": 1794705359000604964,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Set the phone with flat black screen on the angled brown phonestand.",
      "native_instruction": "Set the phone with flat black screen on the angled brown phonestand.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-41-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "09242ce0e0d17ad989c0234543aeba1e8af430ce1664502eaa34b2c3038fa101",
        "source_sha256": "fc1fdd64c38c0c9681a9af78a8b57b36dc15d158bb7795f6d1b38d1cddde5d02",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-42-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/42",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "42",
      "seed": 100001,
      "source_episode_id": "42-place-shoe-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 204.91600973298773,
      "wall_time_s": 1354.038071826,
      "timing": {
        "wall_time_s": 1354.038071826,
        "simulation_time_s": 204.91600973298773,
        "physics_steps": 51229,
        "wall_time_start_ns": 1794704508056198405,
        "wall_time_end_ns": 1794705862094277454,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the medium adult shoe from the table and set it on the mat",
      "native_instruction": "Pick up the medium adult shoe from the table and set it on the mat",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-42-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "13b40e4f60d037c8022d690bf116249201eeb0b571e721af76e0ef2ddeb224be",
        "source_sha256": "26b2e3b99752d19c2d364db8d5b9b42389b728c8b2202a489f349089ed80a034",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-43-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/43",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "43",
      "seed": 100000,
      "source_episode_id": "43-press-stapler-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 177.53600843250751,
      "wall_time_s": 1134.874470857,
      "timing": {
        "wall_time_s": 1134.874470857,
        "simulation_time_s": 177.53600843250751,
        "physics_steps": 44384,
        "wall_time_start_ns": 1794704524404827599,
        "wall_time_end_ns": 1794705659279302516,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Press the plastic stapler firmly using your arm.",
      "native_instruction": "Press the plastic stapler firmly using your arm.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-43-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "9513a91b8557b4d535516e1d3b024f3ea563f3b154bdd1739cda2f4b60d3895e",
        "source_sha256": "abe533c6508580b526fa04f68843cbc3a801da4bbbbbb6ab31c701b27bab286d",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-44-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/44",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "44",
      "seed": 100000,
      "source_episode_id": "44-rotate-qrcode-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 176.38000837760046,
      "wall_time_s": 1059.290468586,
      "timing": {
        "wall_time_s": 1059.290468586,
        "simulation_time_s": 176.38000837760046,
        "physics_steps": 44095,
        "wall_time_start_ns": 1794704490652696890,
        "wall_time_end_ns": 1794705549943169092,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
      "native_instruction": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-44-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "e67421c96c4c3121f9fc6ea3ab1f2fcc50c4332079eeb4a54a0d2bd0d16a9d59",
        "source_sha256": "cc013c3a30dbe4ae493cbe9071efb9294f08095ad9db42ba573823221c5e38c9",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-45-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/45",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "45",
      "seed": 100001,
      "source_episode_id": "45-shake-bottle-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 274.63601304451004,
      "wall_time_s": 1063.469215676,
      "timing": {
        "wall_time_s": 1063.469215676,
        "simulation_time_s": 274.63601304451004,
        "physics_steps": 68659,
        "wall_time_start_ns": 1794704789657230640,
        "wall_time_end_ns": 1794705853126451680,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the smooth bottle with blue text and shake it.",
      "native_instruction": "Pick up the smooth bottle with blue text and shake it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-45-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "6c799d0fb880e9b536b44cac970eb9e8afddf8735bab6501ab2b273a8e463933",
        "source_sha256": "dea997bd22e400e324a726b155c314ef8dcf7d8ac411e3335348594f9af3ed23",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-46-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/46",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "46",
      "seed": 100001,
      "source_episode_id": "46-shake-bottle-horizontally-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 278.3280132198706,
      "wall_time_s": 1493.406525011,
      "timing": {
        "wall_time_s": 1493.406525011,
        "simulation_time_s": 278.3280132198706,
        "physics_steps": 69582,
        "wall_time_start_ns": 1794704845797804430,
        "wall_time_end_ns": 1794706339204335376,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
      "native_instruction": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-46-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "c2137b9ba28eebdab5769b2aacebadc08aa30f185df441539b409c7e6170f7a5",
        "source_sha256": "e4102c4aa1bb95e09896df992c89e485b98c164200ff0527e39f36a5e3858160",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-47-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/47",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "47",
      "seed": 100000,
      "source_episode_id": "47-stack-blocks-two-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 705
      },
      "steps": 705,
      "simulation_time_s": 321.70801528031006,
      "wall_time_s": 1173.156639981,
      "timing": {
        "wall_time_s": 1173.156639981,
        "simulation_time_s": 321.70801528031006,
        "physics_steps": 80427,
        "wall_time_start_ns": 1794764111666498350,
        "wall_time_end_ns": 1794765284823144774,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Move red block to the center, then put green block on top.",
      "native_instruction": "Move red block to the center, then put green block on top.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 705 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-47-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 23,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.65,
        "native_frames": 706,
        "published_frames": 706,
        "decode_ok": true,
        "sha256": "0b2405c8e11046d72f496d176f19cd6ea76cccc0810d1f510d1755bfdddf0c80",
        "source_sha256": "3ced09df4820470942608483894154c19ce81a308df89be76fb646915fb0bafb",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-48-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/48",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "48",
      "seed": 100000,
      "source_episode_id": "48-stack-bowls-two-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 900
      },
      "steps": 900,
      "simulation_time_s": 389.0120184770785,
      "wall_time_s": 1681.312354444,
      "timing": {
        "wall_time_s": 1681.312354444,
        "simulation_time_s": 389.0120184770785,
        "physics_steps": 97253,
        "wall_time_start_ns": 1794764123531894165,
        "wall_time_end_ns": 1794765804844252935,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
      "native_instruction": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 900 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-48-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 29,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 22.5,
        "native_frames": 900,
        "published_frames": 900,
        "decode_ok": true,
        "sha256": "2b1009757c8c564c287d0faa0fa834b4f59fbad62c8737b70c4fc7d4b3171d76",
        "source_sha256": "c3e355d836274a3acdf5347bccdf6de71a7a9d605046668222980ecf7bb031d5",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-49-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/49",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "49",
      "seed": 100003,
      "source_episode_id": "49-stamp-seal-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 197.75600939290598,
      "wall_time_s": 808.090900904,
      "timing": {
        "wall_time_s": 808.090900904,
        "simulation_time_s": 197.75600939290598,
        "physics_steps": 49439,
        "wall_time_start_ns": 1794705032255424203,
        "wall_time_end_ns": 1794705840346329171,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the block-shaped seal with angled edges, press onto Blue",
      "native_instruction": "Grab the block-shaped seal with angled edges, press onto Blue",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-49-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100003,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "0211933e99d8da2834bfd6871970cd646e212c23f2ff6747f9f7f6ad22730818",
        "source_sha256": "b0d4442addf89cddc9fa843d5e2ec68bbc4936819d24bb1df5fb30bf65287a0b",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-image_only-50-20261113",
      "variant": "pose-base-mode1",
      "task_key": "task03/50",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "50",
      "seed": 100000,
      "source_episode_id": "50-turn-switch-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 156.9080074527301,
      "wall_time_s": 1027.311444229,
      "timing": {
        "wall_time_s": 1027.311444229,
        "simulation_time_s": 156.9080074527301,
        "physics_steps": 39227,
        "wall_time_start_ns": 1794705145711715790,
        "wall_time_end_ns": 1794706173023163553,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Press the flat square gray switch using the left arm",
      "native_instruction": "Press the flat square gray switch using the left arm",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-image_only-50-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "image_only",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "9d23d9bab0df664ad6432f86a055d4459859cae1d4e406997fa0e05315d819aa",
        "source_sha256": "b38c8b93810e294b5df1ec4337bdc43fe9288b1c1ac0f6e798731469d9bcea08",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-01-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/01",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "01",
      "seed": 100001,
      "source_episode_id": "01-handover-block-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 800
      },
      "steps": 800,
      "simulation_time_s": 238.6720113363117,
      "wall_time_s": 2048.087594486,
      "timing": {
        "wall_time_s": 2048.087594486,
        "simulation_time_s": 238.6720113363117,
        "physics_steps": 59668,
        "wall_time_start_ns": 1794687690852055551,
        "wall_time_end_ns": 1794689738939655350,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
      "native_instruction": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 800 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-01-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 25,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 20.0,
        "native_frames": 800,
        "published_frames": 800,
        "decode_ok": true,
        "sha256": "7b4ce3c19cb5d62247776afeb143f4986193ea701acee0958acdcd30602b4381",
        "source_sha256": "9b53eeeb6da77e3cd871345c6df37cac40e2e0ea4bd77fbb0360b164b0e017cc",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-02-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/02",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "02",
      "seed": 100001,
      "source_episode_id": "02-place-dual-shoes-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 600
      },
      "steps": 600,
      "simulation_time_s": 165.66400786861777,
      "wall_time_s": 2042.637683198,
      "timing": {
        "wall_time_s": 2042.637683198,
        "simulation_time_s": 165.66400786861777,
        "physics_steps": 41416,
        "wall_time_start_ns": 1794687812477473148,
        "wall_time_end_ns": 1794689855115160761,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
      "native_instruction": "Put a pair of the brown shoe with sturdy sole in the orange box for keeping shoes, tips pointing left.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 600 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-02-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 15.0,
        "native_frames": 600,
        "published_frames": 600,
        "decode_ok": true,
        "sha256": "c78cc24894b206580e73bf822ba2d2399b0a78ccdfb9c6e72821ab8a5f2185b5",
        "source_sha256": "f9445c5251452fb7a78cfb2e98d7f9902d00048b9e33df510abc178acbfd352e",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-03-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/03",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "03",
      "seed": 100000,
      "source_episode_id": "03-put-bottles-dustbin-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1700
      },
      "steps": 1700,
      "simulation_time_s": 537.1760255144909,
      "wall_time_s": 3845.939912686,
      "timing": {
        "wall_time_s": 3845.939912686,
        "simulation_time_s": 537.1760255144909,
        "physics_steps": 134294,
        "wall_time_start_ns": 1794687824286022288,
        "wall_time_end_ns": 1794691670225938170,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
      "native_instruction": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-03-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1700,
        "max_control_steps": 1700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 54,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 42.5,
        "native_frames": 1700,
        "published_frames": 1700,
        "decode_ok": true,
        "sha256": "6801bf41fd67fca294514169ea283455737eef342cf8f1808bb9db52ce0e3a64",
        "source_sha256": "d4c4e9a5301a8d317bd0bcc7a3ebd60ccbeb3e10ae4ea0f73934106af244b41b",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184,
          1216,
          1248,
          1280,
          1312,
          1344,
          1376,
          1408,
          1440,
          1472,
          1504,
          1536,
          1568,
          1600,
          1632,
          1664,
          1696
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-04-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/04",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "04",
      "seed": 100001,
      "source_episode_id": "04-scan-object-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 143.02400679327548,
      "wall_time_s": 1852.719461069,
      "timing": {
        "wall_time_s": 1852.719461069,
        "simulation_time_s": 143.02400679327548,
        "physics_steps": 35756,
        "wall_time_start_ns": 1794687866447745293,
        "wall_time_end_ns": 1794689719167210598,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
      "native_instruction": "Secure both items simultaneously, holding the light gray scanner near the rectangular green box for tea for scanning.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-04-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "f40a74ac69a6d2bf2069f4e18ae34efa1d0dbf35580738934a7b12bd8a407016",
        "source_sha256": "9e7488d11b1be72fdd51cd59985db13fc410e0de5cf548f94c0e08c6fab1afe6",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-05-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/05",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "05",
      "seed": 100000,
      "source_episode_id": "05-stack-bowls-three-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 863
      },
      "steps": 863,
      "simulation_time_s": 271.0120128723793,
      "wall_time_s": 3067.501626374,
      "timing": {
        "wall_time_s": 3067.501626374,
        "simulation_time_s": 271.0120128723793,
        "physics_steps": 67753,
        "wall_time_start_ns": 1794687892941388490,
        "wall_time_end_ns": 1794690960443019596,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
      "native_instruction": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 863 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-05-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 27,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 21.6,
        "native_frames": 864,
        "published_frames": 864,
        "decode_ok": true,
        "sha256": "0498df11c50893d2f205e98693ecb5c9d89e554ef9851adfb1877324a72c70c5",
        "source_sha256": "1fb0c54112074d08f061e46718c31d7b91f0c564e7aa6cd464d2a8744edcf5c0",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-06-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/06",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "06",
      "seed": 100000,
      "source_episode_id": "06-stack-blocks-three-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 368.4600175009109,
      "wall_time_s": 4050.906504081,
      "timing": {
        "wall_time_s": 4050.906504081,
        "simulation_time_s": 368.4600175009109,
        "physics_steps": 92115,
        "wall_time_start_ns": 1794688043403841423,
        "wall_time_end_ns": 1794692094310348448,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
      "native_instruction": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-06-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "798f05273c9ae758383913a47cfedebb3d4948c4f37ce259c1f03df812daba69",
        "source_sha256": "d64fc8b8d4cd79e346db95caecbda91ad84f4f0ffea47999042a106e2a2078b9",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-07-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/07",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "07",
      "seed": 100000,
      "source_episode_id": "07-hanging-mug-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 900
      },
      "steps": 900,
      "simulation_time_s": 252.02801197068766,
      "wall_time_s": 3319.973624633,
      "timing": {
        "wall_time_s": 3319.973624633,
        "simulation_time_s": 252.02801197068766,
        "physics_steps": 63007,
        "wall_time_start_ns": 1794688075558271677,
        "wall_time_end_ns": 1794691395531900146,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
      "native_instruction": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 900 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-07-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 29,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 22.5,
        "native_frames": 900,
        "published_frames": 900,
        "decode_ok": true,
        "sha256": "e0de6e71773d060ed9e342f07fc876cf89652c69a8060bc7574f01fd0f69b550",
        "source_sha256": "bae91b25261eff7940c6e0b0cc0113931426530f0be0d6de65092e184c74e785",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-08-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/08",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "08",
      "seed": 100001,
      "source_episode_id": "08-put-object-cabinet-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 165.38800785550848,
      "wall_time_s": 2276.020998969,
      "timing": {
        "wall_time_s": 2276.020998969,
        "simulation_time_s": 165.38800785550848,
        "physics_steps": 41347,
        "wall_time_start_ns": 1794688406629773036,
        "wall_time_end_ns": 1794690682650776415,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
      "native_instruction": "Pull out the drawer of the cabinet with smooth finish and set the light yellow soap for hygiene into it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-08-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "4777ed0a2c3a23636966e76c6958223dc27ab5892f76a46d3199cad48f22606d",
        "source_sha256": "6c64e022d8ec222820d1f04d6badb6af1fc337c9cd73bbdb88a9e028d74b289f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-09-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/09",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "09",
      "seed": 100000,
      "source_episode_id": "09-blocks-ranking-size-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 449
      },
      "steps": 449,
      "simulation_time_s": 143.26800680486485,
      "wall_time_s": 1400.886244186,
      "timing": {
        "wall_time_s": 1400.886244186,
        "simulation_time_s": 143.26800680486485,
        "physics_steps": 35817,
        "wall_time_start_ns": 1794690067146387746,
        "wall_time_end_ns": 1794691468032635805,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place large block, medium block, and small block in order at the table's center.",
      "native_instruction": "Place large block, medium block, and small block in order at the table's center.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 449 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-09-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 15,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 11.25,
        "native_frames": 450,
        "published_frames": 450,
        "decode_ok": true,
        "sha256": "92de995fe2877413ed39d179ace75d74201379a819f86e67f815bd8dd6911344",
        "source_sha256": "888324325d0cb8a52b7cdfc2e9070ba4f6de653b9826d4ea8778d991fce10213",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-10-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/10",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "10",
      "seed": 100000,
      "source_episode_id": "10-lift-pot-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 96.77200459642336,
      "wall_time_s": 1316.821404166,
      "timing": {
        "wall_time_s": 1316.821404166,
        "simulation_time_s": 96.77200459642336,
        "physics_steps": 24193,
        "wall_time_start_ns": 1794690114622780705,
        "wall_time_end_ns": 1794691431444188580,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the dark gray cooking kitchenpot using both arms",
      "native_instruction": "Lift the dark gray cooking kitchenpot using both arms",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-10-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "2e7758a92545a904ebf2dbbe46329f88b73b8d9f9806c38b0d289d8ed078a7ae",
        "source_sha256": "725bc576f6aab4ad75a6437393e552d33c43b7f66f9132e609f0a23d70a36d37",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-11-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/11",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "11",
      "seed": 100000,
      "source_episode_id": "11-adjust-bottle-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 138.4240065747872,
      "wall_time_s": 1180.350328809,
      "timing": {
        "wall_time_s": 1180.350328809,
        "simulation_time_s": 138.4240065747872,
        "physics_steps": 34606,
        "wall_time_start_ns": 1794690236352033850,
        "wall_time_end_ns": 1794691416702366128,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the brown bottle upright",
      "native_instruction": "Use the left arm to grab the brown bottle upright",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-11-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "a25543e5986ed4f8da9b3e22460353ea8e73ee8be0b2e259dcf248e50c255d60",
        "source_sha256": "e2949c10eba29b1b71b77c6730d70e402beb566e2837b6fafa2aefb7165317f6",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-12-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/12",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "12",
      "seed": 100000,
      "source_episode_id": "12-beat-block-hammer-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 96
      },
      "steps": 96,
      "simulation_time_s": 32.4480015411973,
      "wall_time_s": 132.990736921,
      "timing": {
        "wall_time_s": 132.990736921,
        "simulation_time_s": 32.4480015411973,
        "physics_steps": 8112,
        "wall_time_start_ns": 1794690281168169188,
        "wall_time_end_ns": 1794690414158911401,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the hammer with black handle then beat block",
      "native_instruction": "Use the left arm to grab the hammer with black handle then beat block",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 96 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-12-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 2.425,
        "native_frames": 97,
        "published_frames": 97,
        "decode_ok": true,
        "sha256": "4046fa3c4dc028a40fb25fe2909595997c7d22ca56100dd32747e8f961797b9f",
        "source_sha256": "d8f2e92f985ccc1a72825d06c0ba59fe2b33a7839cc820148ff74babcf524856",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-13-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/13",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "13",
      "seed": 100000,
      "source_episode_id": "13-blocks-ranking-rgb-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 449.8400213662535,
      "wall_time_s": 1679.481660118,
      "timing": {
        "wall_time_s": 1679.481660118,
        "simulation_time_s": 449.8400213662535,
        "physics_steps": 112460,
        "wall_time_start_ns": 1794690306023683201,
        "wall_time_end_ns": 1794691985505345996,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
      "native_instruction": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1200 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-13-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "00c654046772a29a35938389361ffeda38e488b4cda3b08def923328ee41fe9c",
        "source_sha256": "1a03e73119c7273183d605542d39fa50ad851667bbaec54f6adbc49a49040afd",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-14-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/14",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "14",
      "seed": 100000,
      "source_episode_id": "14-click-alarmclock-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 62
      },
      "steps": 62,
      "simulation_time_s": 20.892000992316753,
      "wall_time_s": 141.53591375,
      "timing": {
        "wall_time_s": 141.53591375,
        "simulation_time_s": 20.892000992316753,
        "physics_steps": 5223,
        "wall_time_start_ns": 1794690416469217871,
        "wall_time_end_ns": 1794690558005135771,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Tap the center top button of the alarm-clock with beige screen",
      "native_instruction": "Tap the center top button of the alarm-clock with beige screen",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 62 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-14-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 2,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.575,
        "native_frames": 63,
        "published_frames": 63,
        "decode_ok": true,
        "sha256": "e79eb25e1596c19f6830485b184ee4d1b5968fd3143b1bf7f0e4ad509abb05ad",
        "source_sha256": "1ed3294d574b592f209e3e33d4306d111286b001db0a5aabcedfdd7be96e5f5d",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-15-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/15",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "15",
      "seed": 100000,
      "source_episode_id": "15-click-bell-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 52
      },
      "steps": 52,
      "simulation_time_s": 20.668000981677324,
      "wall_time_s": 123.053164077,
      "timing": {
        "wall_time_s": 123.053164077,
        "simulation_time_s": 20.668000981677324,
        "physics_steps": 5167,
        "wall_time_start_ns": 1794690459378528908,
        "wall_time_end_ns": 1794690582431698658,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to press the medium-sized bell's top center.",
      "native_instruction": "Use the left arm to press the medium-sized bell's top center.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 52 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-15-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 2,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.325,
        "native_frames": 53,
        "published_frames": 53,
        "decode_ok": true,
        "sha256": "eabdb1843505463305808c4351594827f9e966fe75e7640c9806e9c0cf7035c0",
        "source_sha256": "667f26ae4e8353dc9162524d0899d7ee1e4cc63cc12f0d3fd9cb44d8a745c0b3",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-16-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/16",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "16",
      "seed": 100006,
      "source_episode_id": "16-dump-bin-bigbin-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 600
      },
      "steps": 600,
      "simulation_time_s": 204.50800971360877,
      "wall_time_s": 1463.025889299,
      "timing": {
        "wall_time_s": 1463.025889299,
        "simulation_time_s": 204.50800971360877,
        "physics_steps": 51127,
        "wall_time_start_ns": 1794698081477709838,
        "wall_time_end_ns": 1794699544503602801,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
      "native_instruction": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 600 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-16-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100006,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 15.0,
        "native_frames": 600,
        "published_frames": 600,
        "decode_ok": true,
        "sha256": "b5b5fd7222583b87691f1082b1335772539eea24fd88aa7021402680d3a3308c",
        "source_sha256": "6a9d66cd404dc874a26e78329a4778e79fe5fb665d7fd6c8aea328722ad0fab7",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-17-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/17",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "17",
      "seed": 100000,
      "source_episode_id": "17-grab-roller-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 76
      },
      "steps": 76,
      "simulation_time_s": 29.080001381225884,
      "wall_time_s": 158.537827879,
      "timing": {
        "wall_time_s": 158.537827879,
        "simulation_time_s": 29.080001381225884,
        "physics_steps": 7270,
        "wall_time_start_ns": 1794698116134414148,
        "wall_time_end_ns": 1794698274672246622,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use arms to grab the smooth wooden roller.",
      "native_instruction": "Use arms to grab the smooth wooden roller.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 76 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-17-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 1.925,
        "native_frames": 77,
        "published_frames": 77,
        "decode_ok": true,
        "sha256": "357c640657b5a33d1cb67c0ed555c437c04c0bad65b29e5f059c5f41b18224c7",
        "source_sha256": "ce0f2108e824290523d2ca38809242e9b250157d2acbda755b9cb12cb735a0d9",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-18-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/18",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "18",
      "seed": 100000,
      "source_episode_id": "18-handover-mic-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 600
      },
      "steps": 600,
      "simulation_time_s": 145.70400692056865,
      "wall_time_s": 1386.707104946,
      "timing": {
        "wall_time_s": 1386.707104946,
        "simulation_time_s": 145.70400692056865,
        "physics_steps": 36426,
        "wall_time_start_ns": 1794698125584766423,
        "wall_time_end_ns": 1794699512291876041,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use one arm to hold the microphone used for audio input and hand it over.",
      "native_instruction": "Use one arm to hold the microphone used for audio input and hand it over.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 600 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-18-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 15.0,
        "native_frames": 600,
        "published_frames": 600,
        "decode_ok": true,
        "sha256": "1a3190066b6dcc0962fa0d689e40f5fd0ccbee44d821e8898a662978ca83e655",
        "source_sha256": "08a9077d26b91dd02b236743483ce333d9a4e3924c400e33ab0c13062234c48c",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-19-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/19",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "19",
      "seed": 100000,
      "source_episode_id": "19-move-can-pot-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 121.2760057603009,
      "wall_time_s": 1239.061343953,
      "timing": {
        "wall_time_s": 1239.061343953,
        "simulation_time_s": 121.2760057603009,
        "physics_steps": 30319,
        "wall_time_start_ns": 1794698249143318134,
        "wall_time_end_ns": 1794699488204666783,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
      "native_instruction": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-19-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "3c7f6974df8a85c501d3e37914c3a36794047442394b99e06355eaf047ae477b",
        "source_sha256": "8940e33dadd2e7fba55408d1527dc49c02184cc92c446531fffd47a7c872d445",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-20-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/20",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "20",
      "seed": 100001,
      "source_episode_id": "20-move-pillbottle-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 134.01600636541843,
      "wall_time_s": 1166.985109597,
      "timing": {
        "wall_time_s": 1166.985109597,
        "simulation_time_s": 134.01600636541843,
        "physics_steps": 33504,
        "wall_time_start_ns": 1794698286556928274,
        "wall_time_end_ns": 1794699453542042184,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the white cylindrical bottle and set it on the pad",
      "native_instruction": "Grab the white cylindrical bottle and set it on the pad",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-20-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "e8a5e9fe30c3fe3658ed78bec12cf200b12452c6b82bcfd15bee211119b664e0",
        "source_sha256": "ea9bba498e6c1f3cb58552f88cde5a526ce261fbeed10a7d58c1e8aadf803343",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-21-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/21",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "21",
      "seed": 100000,
      "source_episode_id": "21-move-playingcard-away-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 107.92800512630492,
      "wall_time_s": 1185.725929599,
      "timing": {
        "wall_time_s": 1185.725929599,
        "simulation_time_s": 107.92800512630492,
        "physics_steps": 26982,
        "wall_time_start_ns": 1794698446429250808,
        "wall_time_end_ns": 1794699632155189142,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the right arm to grab the hard black box for card deck and relocate it.",
      "native_instruction": "Use the right arm to grab the hard black box for card deck and relocate it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-21-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "c2e044bb06b5ffa10d2fff60d9488d019173427798542b846c8a18f7cc2eed27",
        "source_sha256": "9ffec81152457cd725c7becfa3ec235025e802cd11d3a76319fb6d09b8e4a2a2",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-22-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/22",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "22",
      "seed": 100000,
      "source_episode_id": "22-move-stapler-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 116.22400552034378,
      "wall_time_s": 1160.098872569,
      "timing": {
        "wall_time_s": 1160.098872569,
        "simulation_time_s": 116.22400552034378,
        "physics_steps": 29056,
        "wall_time_start_ns": 1794698677934815997,
        "wall_time_end_ns": 1794699838033694993,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to move the plastic stapler to Yellow mat",
      "native_instruction": "Use the left arm to move the plastic stapler to Yellow mat",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-22-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "291ac44f5a3f9206e9bd0e5e170b3aba79715f63c64a07ae2a26adb480b88c64",
        "source_sha256": "c49fd5b47e25d2446f9368022debe68bc45623437ad4b277db1bf195b1b663ff",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-23-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/23",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "23",
      "seed": 100000,
      "source_episode_id": "23-open-laptop-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 166.6000079130754,
      "wall_time_s": 2328.35315479,
      "timing": {
        "wall_time_s": 2328.35315479,
        "simulation_time_s": 166.6000079130754,
        "physics_steps": 41650,
        "wall_time_start_ns": 1794699791211174972,
        "wall_time_end_ns": 1794702119564333007,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Open the flat orange laptop with one motion.",
      "native_instruction": "Open the flat orange laptop with one motion.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-23-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "69c3f8e86d552ed32c7a72df36645dbd0836ca9d99add3c2dd666e38133ea9a7",
        "source_sha256": "a927568124ea3bf23a55a9e4b39d27dac5e256694f6a22a69c605162d05ece1a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-24-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/24",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "24",
      "seed": 100000,
      "source_episode_id": "24-open-microwave-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1500
      },
      "steps": 1500,
      "simulation_time_s": 399.19601896079257,
      "wall_time_s": 3829.788500113,
      "timing": {
        "wall_time_s": 3829.788500113,
        "simulation_time_s": 399.19601896079257,
        "physics_steps": 99799,
        "wall_time_start_ns": 1794699698384966088,
        "wall_time_end_ns": 1794703528173469955,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
      "native_instruction": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 1500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-24-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1500,
        "max_control_steps": 1500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 47,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 37.5,
        "native_frames": 1500,
        "published_frames": 1500,
        "decode_ok": true,
        "sha256": "78dbe61e411886be1bca026adc59c9ed6670908a0f87f7d974c96976911eceaa",
        "source_sha256": "950852af1329e08504abb1f66aab43dfb2476abee2ad9c19685ee8e356d13091",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184,
          1216,
          1248,
          1280,
          1312,
          1344,
          1376,
          1408,
          1440,
          1472
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-25-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/25",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "25",
      "seed": 100005,
      "source_episode_id": "25-pick-diverse-bottles-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 92.81600440852344,
      "wall_time_s": 1077.588950791,
      "timing": {
        "wall_time_s": 1077.588950791,
        "simulation_time_s": 92.81600440852344,
        "physics_steps": 23204,
        "wall_time_start_ns": 1794699760974036889,
        "wall_time_end_ns": 1794700838562991946,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
      "native_instruction": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-25-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100005,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "04a70874cacce3a649762ad8c6335b65fbe02711e1c91611fc58f4f33ade42c8",
        "source_sha256": "0f3f3a41e3b3e0cf53639115bcaf72f2f2ce2d42a13e26cea514554abb6bed77",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-26-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/26",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "26",
      "seed": 100000,
      "source_episode_id": "26-pick-dual-bottles-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 92.76400440605357,
      "wall_time_s": 1080.866169694,
      "timing": {
        "wall_time_s": 1080.866169694,
        "simulation_time_s": 92.76400440605357,
        "physics_steps": 23191,
        "wall_time_start_ns": 1794699787673786287,
        "wall_time_end_ns": 1794700868539959898,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
      "native_instruction": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-26-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "d3e68d46614e46dc9c88db2ac99e8b7087a0fc106b68a89fa32e7a2672eefe3e",
        "source_sha256": "af60c543b98067f1e62f89528a0960f17cc0796b3fa1f6b760b457c2b7a405d4",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-27-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/27",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "27",
      "seed": 100000,
      "source_episode_id": "27-place-a2b-left-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 126
      },
      "steps": 126,
      "simulation_time_s": 38.96800185088068,
      "wall_time_s": 421.952312999,
      "timing": {
        "wall_time_s": 421.952312999,
        "simulation_time_s": 38.96800185088068,
        "physics_steps": 9742,
        "wall_time_start_ns": 1794699983046728697,
        "wall_time_end_ns": 1794700404999047303,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
      "native_instruction": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 126 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-27-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 3.175,
        "native_frames": 127,
        "published_frames": 127,
        "decode_ok": true,
        "sha256": "e97c2c9984a3f943ae16fcee405c1dc62762a2aae13b71ef036c6c0b1b29610f",
        "source_sha256": "862bc7064cefc4d5bc8769e7cf7edb1e6dd0faa7454d59b4ecc551d30e597663",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-28-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/28",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "28",
      "seed": 100000,
      "source_episode_id": "28-place-a2b-right-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 121.32000576239079,
      "wall_time_s": 1127.132285248,
      "timing": {
        "wall_time_s": 1127.132285248,
        "simulation_time_s": 121.32000576239079,
        "physics_steps": 30330,
        "wall_time_start_ns": 1794700133993565589,
        "wall_time_end_ns": 1794701261125855111,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
      "native_instruction": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-28-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "d956a7ae33693462f3e0d75d49f51c629e6fff86c74ba514bfb661cb6a199ffc",
        "source_sha256": "687bece2fd3218ba99145356264a9bd97674b6d7bfc86067258733f104871361",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-29-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/29",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "29",
      "seed": 100000,
      "source_episode_id": "29-place-bread-basket-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 243.7680115783587,
      "wall_time_s": 2191.885996337,
      "timing": {
        "wall_time_s": 2191.885996337,
        "simulation_time_s": 243.7680115783587,
        "physics_steps": 60942,
        "wall_time_start_ns": 1794700696692741794,
        "wall_time_end_ns": 1794702888578742286,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
      "native_instruction": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-29-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "82b631f6f726757fcdc0dc0e5e83287ca87bbf4b91c56ad045a4ba930e189301",
        "source_sha256": "e488cb5bf1637133556c7248c9c73570a11728d800e88cc370bbf067db6e139f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-30-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/30",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "30",
      "seed": 100001,
      "source_episode_id": "30-place-bread-skillet-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 130.21200618473813,
      "wall_time_s": 1695.487729218,
      "timing": {
        "wall_time_s": 1695.487729218,
        "simulation_time_s": 130.21200618473813,
        "physics_steps": 32553,
        "wall_time_start_ns": 1794701340666479806,
        "wall_time_end_ns": 1794703036154213207,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
      "native_instruction": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-30-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "faa0db132256a227c463ec038b615f6c0f2f1bde90dd421696bea783316dd4ce",
        "source_sha256": "75074a87f09e1237996e40fe1f77b94821fd66dd3d81abd1ab6ca70ef7a548ea",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-31-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/31",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "31",
      "seed": 100000,
      "source_episode_id": "31-place-burger-fries-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 151.70400720555335,
      "wall_time_s": 1779.780699682,
      "timing": {
        "wall_time_s": 1779.780699682,
        "simulation_time_s": 151.70400720555335,
        "physics_steps": 37926,
        "wall_time_start_ns": 1794701478829251385,
        "wall_time_end_ns": 1794703258609954730,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
      "native_instruction": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-31-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "03772f63a9752b5267783b1bcf5425462ade7f30a26f29653553ec94fa350fa4",
        "source_sha256": "dccd6c05039a35f3dbe70168ee38514192b98a9a8c579717805a627f0a8a952a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-32-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/32",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "32",
      "seed": 100000,
      "source_episode_id": "32-place-can-basket-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 230.10401092935354,
      "wall_time_s": 2507.761782802,
      "timing": {
        "wall_time_s": 2507.761782802,
        "simulation_time_s": 230.10401092935354,
        "physics_steps": 57526,
        "wall_time_start_ns": 1794701767668223455,
        "wall_time_end_ns": 1794704275430010935,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "native_instruction": "Use the left arm to place the palm-sized red metal Coca-Cola can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-32-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "9d25069161b1f8b7468a0185af8c8c45751858aefe0f1837ee79456b6b6aa819",
        "source_sha256": "40bf634586b677b0d15b89e5ce3dc2bd5f1f4c7fd2a3c00aa76f6a771cb12946",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-33-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/33",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "33",
      "seed": 100000,
      "source_episode_id": "33-place-cans-plasticbox-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 800
      },
      "steps": 800,
      "simulation_time_s": 255.9400121564977,
      "wall_time_s": 2586.64453611,
      "timing": {
        "wall_time_s": 2586.64453611,
        "simulation_time_s": 255.9400121564977,
        "physics_steps": 63985,
        "wall_time_start_ns": 1794702547301421388,
        "wall_time_end_ns": 1794705133945963562,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
      "native_instruction": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 800 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-33-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 25,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 20.0,
        "native_frames": 800,
        "published_frames": 800,
        "decode_ok": true,
        "sha256": "a0a58054f00050bad11a71895c5801be8064500bf2c2e379716299abaec121f7",
        "source_sha256": "287b676d35a645a9d288e01476456fd1f0a596325fdb8d1ccf51b3e1bdf7410c",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-34-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/34",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "34",
      "seed": 100000,
      "source_episode_id": "34-place-container-plate-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 126
      },
      "steps": 126,
      "simulation_time_s": 50.592002402991056,
      "wall_time_s": 419.917541973,
      "timing": {
        "wall_time_s": 419.917541973,
        "simulation_time_s": 50.592002402991056,
        "physics_steps": 12648,
        "wall_time_start_ns": 1794703183803220061,
        "wall_time_end_ns": 1794703603720767408,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the curved bowl with smooth texture on the gray plate for dining.",
      "native_instruction": "Place the curved bowl with smooth texture on the gray plate for dining.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 126 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-34-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 3.175,
        "native_frames": 127,
        "published_frames": 127,
        "decode_ok": true,
        "sha256": "ece928bef7915f6d37c206b5b48b4082f55a013b9c878835a80cb87527eaf105",
        "source_sha256": "b0f2e912cdc8bc25f44d261c35a734f8e6ec36578b63afd0e60849b9427e6872",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-35-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/35",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "35",
      "seed": 100000,
      "source_episode_id": "35-place-empty-cup-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 146.12400694051757,
      "wall_time_s": 1564.502945633,
      "timing": {
        "wall_time_s": 1564.502945633,
        "simulation_time_s": 146.12400694051757,
        "physics_steps": 36531,
        "wall_time_start_ns": 1794703428464307464,
        "wall_time_end_ns": 1794704992967260583,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
      "native_instruction": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-35-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "307013db3060bd98953af51632da0ca3491af508697c475c171ef2d35a22e2f2",
        "source_sha256": "00e8d13df0781c05ff26bc828220878941322a684b93315db468062191ec6541",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-36-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/36",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "36",
      "seed": 100000,
      "source_episode_id": "36-place-fan-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 137.26400651969016,
      "wall_time_s": 1298.142066826,
      "timing": {
        "wall_time_s": 1298.142066826,
        "simulation_time_s": 137.26400651969016,
        "physics_steps": 34316,
        "wall_time_start_ns": 1794703594291842389,
        "wall_time_end_ns": 1794704892433912962,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
      "native_instruction": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-36-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "e186c070ead4273b0ca0323412c4ab6d9cbd75cc42f12e7664681101c6fbde37",
        "source_sha256": "10a314a91f205943d5463dde16785fd65511327119b69f740e159d7a5d327be4",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-37-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/37",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "37",
      "seed": 100000,
      "source_episode_id": "37-place-mouse-pad-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 114.76800545118749,
      "wall_time_s": 1291.165092535,
      "timing": {
        "wall_time_s": 1291.165092535,
        "simulation_time_s": 114.76800545118749,
        "physics_steps": 28692,
        "wall_time_start_ns": 1794703897739055594,
        "wall_time_end_ns": 1794705188904159442,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
      "native_instruction": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-37-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "6d64e8a94df48b00580c720271f81b60496c1def4e05ae22f00e3eeec466e0f6",
        "source_sha256": "4a6ef7f36dcd3d5eea1a51ab5e0c01147a3436e77fb1eca4c147980578a294d4",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-38-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/38",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "38",
      "seed": 100000,
      "source_episode_id": "38-place-object-basket-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 244.96001163497567,
      "wall_time_s": 1644.407730938,
      "timing": {
        "wall_time_s": 1644.407730938,
        "simulation_time_s": 244.96001163497567,
        "physics_steps": 61240,
        "wall_time_start_ns": 1794704090313166739,
        "wall_time_end_ns": 1794705734720902306,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
      "native_instruction": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-38-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "efd3bc4877787ad02303e815b72b11c3f327558aaf26e847344b5ea567994c70",
        "source_sha256": "41d86e74ee90354e95036c6a86fbf9d6ad0052eb70871ecf4c21cd0b0babf4d9",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-39-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/39",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "39",
      "seed": 100000,
      "source_episode_id": "39-place-object-scale-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 126.16000599227846,
      "wall_time_s": 913.455409184,
      "timing": {
        "wall_time_s": 913.455409184,
        "simulation_time_s": 126.16000599227846,
        "physics_steps": 31540,
        "wall_time_start_ns": 1794704316142166245,
        "wall_time_end_ns": 1794705229597582419,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
      "native_instruction": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-39-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "23119a0a5810810fcac788ce878492d3b98c50e926a925b988c9f74bae2d7c97",
        "source_sha256": "a4890f1c6091346d684a6a2218c98a5e71ba941df5d41e1d37a1ab03feece28c",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-40-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/40",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "40",
      "seed": 100001,
      "source_episode_id": "40-place-object-stand-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 118
      },
      "steps": 118,
      "simulation_time_s": 37.94400180224329,
      "wall_time_s": 176.552215483,
      "timing": {
        "wall_time_s": 176.552215483,
        "simulation_time_s": 37.94400180224329,
        "physics_steps": 9486,
        "wall_time_start_ns": 1794704328201518815,
        "wall_time_end_ns": 1794704504753741467,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the white stapler on the compact black hollow displaystand with the left arm.",
      "native_instruction": "Place the white stapler on the compact black hollow displaystand with the left arm.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 118 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-40-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 2.975,
        "native_frames": 119,
        "published_frames": 119,
        "decode_ok": true,
        "sha256": "18eb99b70fc3672dabb6e39e50cd3b55ebe66d4e769e7ebf8cee200184ff2002",
        "source_sha256": "8191962920a9eb1881081e298d121211e83aca1abd8a5b7b5080d3ed09ccae0a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-41-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/41",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "41",
      "seed": 100000,
      "source_episode_id": "41-place-phone-stand-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 129.73600616212934,
      "wall_time_s": 979.023411761,
      "timing": {
        "wall_time_s": 979.023411761,
        "simulation_time_s": 129.73600616212934,
        "physics_steps": 32434,
        "wall_time_start_ns": 1794704345959353219,
        "wall_time_end_ns": 1794705324982769343,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Set the phone with flat black screen on the angled brown phonestand.",
      "native_instruction": "Set the phone with flat black screen on the angled brown phonestand.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-41-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "28c8b1ba886e2a4bba71f75f2c346fd135fa0253b05ee9a09ff4404d60d608db",
        "source_sha256": "11b37a2db514df816d72bb4d079654b7a5a5a597b9617d9320d3924ca43218f4",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-42-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/42",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "42",
      "seed": 100001,
      "source_episode_id": "42-place-shoe-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 136.47600648226216,
      "wall_time_s": 1129.907597442,
      "timing": {
        "wall_time_s": 1129.907597442,
        "simulation_time_s": 136.47600648226216,
        "physics_steps": 34119,
        "wall_time_start_ns": 1794704452814761700,
        "wall_time_end_ns": 1794705582722362881,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the medium adult shoe from the table and set it on the mat",
      "native_instruction": "Pick up the medium adult shoe from the table and set it on the mat",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 500 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-42-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "5f8cd326384e7dc5ac2f5ee425900a658784dcb59e43bb6fd7598f5c76704e0a",
        "source_sha256": "0f490a0e26a86c5eda18f34ee4ed13235161d14229c7236eab5b4e33c6acb037",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-43-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/43",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "43",
      "seed": 100000,
      "source_episode_id": "43-press-stapler-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 180
      },
      "steps": 180,
      "simulation_time_s": 62.128002950921655,
      "wall_time_s": 636.974936681,
      "timing": {
        "wall_time_s": 636.974936681,
        "simulation_time_s": 62.128002950921655,
        "physics_steps": 15532,
        "wall_time_start_ns": 1794704567089678968,
        "wall_time_end_ns": 1794705204064621532,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Press the plastic stapler firmly using your arm.",
      "native_instruction": "Press the plastic stapler firmly using your arm.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 180 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-43-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 6,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 4.525,
        "native_frames": 181,
        "published_frames": 181,
        "decode_ok": true,
        "sha256": "e302822a3e0e4b2834201064585fcf8c0e8138fb03c63507d1197fd947aa28dd",
        "source_sha256": "494879865eaee21b66c256fa2a232d6618ad7e3eb53837e5b7bc0aa93d7c8be3",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-44-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/44",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "44",
      "seed": 100000,
      "source_episode_id": "44-rotate-qrcode-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 124.81600592844188,
      "wall_time_s": 1141.376589699,
      "timing": {
        "wall_time_s": 1141.376589699,
        "simulation_time_s": 124.81600592844188,
        "physics_steps": 31204,
        "wall_time_start_ns": 1794704553501576298,
        "wall_time_end_ns": 1794705694878170716,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
      "native_instruction": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-44-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "a9393c769215c81afee6c979ba6b0492737538e693389b3367202a80e9971f10",
        "source_sha256": "af0411be4c2c6964504b052c1e6cfb81ef51613b177a06df8e777cc13ddf4a76",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-45-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/45",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "45",
      "seed": 100001,
      "source_episode_id": "45-shake-bottle-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 209.77200996363536,
      "wall_time_s": 1141.475966436,
      "timing": {
        "wall_time_s": 1141.475966436,
        "simulation_time_s": 209.77200996363536,
        "physics_steps": 52443,
        "wall_time_start_ns": 1794704811281383791,
        "wall_time_end_ns": 1794705952757353672,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Pick up the smooth bottle with blue text and shake it.",
      "native_instruction": "Pick up the smooth bottle with blue text and shake it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 700 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-45-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "3761437b5725f636dff9b82e5f9572631a173018bdf812b97b7615955dff9bff",
        "source_sha256": "a5c1fa258b65a87ae6c88c29ca7d410d020328221501e88da9b88ebce72f9fb7",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-46-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/46",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "46",
      "seed": 100001,
      "source_episode_id": "46-shake-bottle-horizontally-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 79
      },
      "steps": 79,
      "simulation_time_s": 30.16400143271312,
      "wall_time_s": 349.8673137,
      "timing": {
        "wall_time_s": 349.8673137,
        "simulation_time_s": 30.16400143271312,
        "physics_steps": 7541,
        "wall_time_start_ns": 1794704868880761365,
        "wall_time_end_ns": 1794705218748081279,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
      "native_instruction": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 79 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-46-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 2.0,
        "native_frames": 80,
        "published_frames": 80,
        "decode_ok": true,
        "sha256": "7f88d6628f47609e6575ff34a55135b4801139350ec6e6e99b2a455607337739",
        "source_sha256": "6101b4d46643e84d037ca258e521be00df90f9c4435c026a5b7660eb6d9817ca",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-47-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/47",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "47",
      "seed": 100000,
      "source_episode_id": "47-stack-blocks-two-hard",
      "status": "completed",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 279
      },
      "steps": 279,
      "simulation_time_s": 92.19200437888503,
      "wall_time_s": 182.374560524,
      "timing": {
        "wall_time_s": 182.374560524,
        "simulation_time_s": 92.19200437888503,
        "physics_steps": 23048,
        "wall_time_start_ns": 1794764111500283767,
        "wall_time_end_ns": 1794764293874853533,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Move red block to the center, then put green block on top.",
      "native_instruction": "Move red block to the center, then put green block on top.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 279 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-47-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 9,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 7.0,
        "native_frames": 280,
        "published_frames": 280,
        "decode_ok": true,
        "sha256": "2a6e82b2c0cfceff11f81de4ff82ab7eb36a57747d7d6ac8b7c2b1ab5ef1f3ca",
        "source_sha256": "d5461d9cdc1cb312bf264addec20894f14e304a9475ee9af3670fdbeb1818a2b",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-48-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/48",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "48",
      "seed": 100000,
      "source_episode_id": "48-stack-bowls-two-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 900
      },
      "steps": 900,
      "simulation_time_s": 260.2120123594068,
      "wall_time_s": 1341.552094461,
      "timing": {
        "wall_time_s": 1341.552094461,
        "simulation_time_s": 260.2120123594068,
        "physics_steps": 65053,
        "wall_time_start_ns": 1794705116776137745,
        "wall_time_end_ns": 1794706458328240384,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
      "native_instruction": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 900 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-48-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 29,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 22.5,
        "native_frames": 900,
        "published_frames": 900,
        "decode_ok": true,
        "sha256": "67bf5044ee1150153a95625ecdaf8b38bf57ddee59623740b5a49e487efa6344",
        "source_sha256": "d15069b9933ce8ac15d3e43fa5699bfd7fcaff556bc70f50f50222e081251006",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-49-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/49",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "49",
      "seed": 100003,
      "source_episode_id": "49-stamp-seal-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 138.8560065953061,
      "wall_time_s": 341.365524266,
      "timing": {
        "wall_time_s": 341.365524266,
        "simulation_time_s": 138.8560065953061,
        "physics_steps": 34714,
        "wall_time_start_ns": 1794705210458120971,
        "wall_time_end_ns": 1794705551823649171,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Grab the block-shaped seal with angled edges, press onto Blue",
      "native_instruction": "Grab the block-shaped seal with angled edges, press onto Blue",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-49-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100003,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "90ddb01f92cd9f6a6959105d883c795d1025f959237d31c4f101532994d7e860",
        "source_sha256": "9f243bce5758c4bcbb205b65ed050604b006a6243bb111e33a43220ba9605f4b",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "hard-native-pose_base-reference_pose-50-20261113",
      "variant": "pose-base-mode2",
      "task_key": "task03/50",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "50",
      "seed": 100000,
      "source_episode_id": "50-turn-switch-hard",
      "status": "completed",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 99.61200473131612,
      "wall_time_s": 693.926626457,
      "timing": {
        "wall_time_s": 693.926626457,
        "simulation_time_s": 99.61200473131612,
        "physics_steps": 24903,
        "wall_time_start_ns": 1794705204236751551,
        "wall_time_end_ns": 1794705898163380834,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "progVLA-yz",
      "effort": "50k · native",
      "harness": "standalone",
      "instruction": "Press the flat square gray switch using the left arm",
      "native_instruction": "Press the flat square gray switch using the left arm",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native step-limit failure after 400 official actions. Native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/presentation/video.mp4",
        "poster": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/presentation/poster.jpg",
        "terminal_frame": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/presentation/terminal.jpg",
        "transcript": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/transcript.json",
        "usage": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/usage.json",
        "analysis": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/analysis.json",
        "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/protocol.json",
        "native_goal": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/instruction.json",
        "verdict": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/episode.json",
        "media_check": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/presentation/media-validation.json",
        "recording_manifest": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/presentation/media-validation.json",
        "tool_calls": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/tool-calls.json",
        "artifact_index": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/artifact-index.json",
        "vla_io": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/vla-io.json",
        "evidence_archive": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/evidence.zip",
        "timing": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/timing.json",
        "provenance": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/research/pose-standalone-native-hard-20261113/source-identities.json",
        "native_video": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/native-head.mp4",
        "policy_trace": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/dc0784a3aecd17830694e0e2653ccdcc614ce2d9/experiments/pose-standalone-native-hard-20261113/hard-native-pose_base-reference_pose-50-20261113/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "frozen successful reference certificate; exact initial scene verification",
        "model_seed": 42,
        "official_seed_group": 0,
        "action_chunk_size": 32,
        "pose_conditioning": "reference_pose",
        "progress_head": false
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 960,
        "height": 240,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "8f2f11c57c3d539b30924ea8d43bb75b442f9fe6535e008e401f1c07c34771ae",
        "source_sha256": "9b3c37f9a9a08ce11b68c68f086423909c02aaa651aa3cb63debfae57655e86c",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Exact inference observations held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view; no upscaling",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs can include a post-action head frame; action-limit failures end on the last pre-action head frame. No terminal frames synthesized."
      }
    },
    {
      "id": "native-kinex-01-20261113",
      "variant": "kinex-native",
      "task_key": "task03/01",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "01",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 30.16,
        "step_limit": 800,
        "steps": 754,
        "success": true,
        "valid": true
      },
      "steps": 754,
      "simulation_time_s": 30.16,
      "wall_time_s": 2754.464723,
      "timing": {
        "simulation_time_s": 30.16,
        "wall_time_s": 2754.464723,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "native_instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9790102976618986,
        "cache_read_input_tokens": 2502272,
        "cache_reported_input_tokens": 2555920,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2555920,
        "cached_input_tokens": 2502272,
        "cost_usd": null,
        "fresh_input_tokens": 53648,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2555920,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2502272,
        "known_fresh_input_tokens": 53648,
        "known_input_tokens": 2555920,
        "known_output_tokens": 11275,
        "known_reasoning_output_tokens": 3668,
        "output_tokens": 11275,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3668,
        "reasoning_reported_output_tokens": 11275,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 69,
          "cache_write_input_tokens": 69,
          "cache_write_reported_input_tokens": 69,
          "cached_input_tokens": 69,
          "fresh_input_tokens": 69,
          "input_tokens": 69,
          "output_tokens": 69,
          "reasoning_output_tokens": 69,
          "reasoning_reported_output_tokens": 69
        },
        "request_attempts": 69,
        "response_count": 69,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2567195,
        "uncached_input_tokens": 53648,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11827,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11862,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11993,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12041,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12289,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 16138,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 18668,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 20799,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 20991,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21243,
            "output_tokens": 179,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 22356,
            "output_tokens": 104,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22829,
            "output_tokens": 319,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23525,
            "output_tokens": 404,
            "reasoning_output_tokens": 362
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 25108,
            "output_tokens": 1131,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26260,
            "output_tokens": 43,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26976,
            "output_tokens": 284,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27408,
            "output_tokens": 260,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28353,
            "output_tokens": 93,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 29141,
            "output_tokens": 324,
            "reasoning_output_tokens": 221
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29964,
            "output_tokens": 146,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30606,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 31159,
            "output_tokens": 247,
            "reasoning_output_tokens": 106
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 32218,
            "output_tokens": 145,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 33964,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34324,
            "output_tokens": 124,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34935,
            "output_tokens": 49,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35152,
            "output_tokens": 310,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35608,
            "output_tokens": 151,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36245,
            "output_tokens": 84,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36391,
            "output_tokens": 186,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 37058,
            "output_tokens": 186,
            "reasoning_output_tokens": 102
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37770,
            "output_tokens": 114,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38051,
            "output_tokens": 275,
            "reasoning_output_tokens": 160
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38353,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38618,
            "output_tokens": 149,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39249,
            "output_tokens": 109,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 40125,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40216,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40591,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40782,
            "output_tokens": 163,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41315,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41732,
            "output_tokens": 71,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42174,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 43076,
            "output_tokens": 903,
            "reasoning_output_tokens": 739
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44666,
            "output_tokens": 337,
            "reasoning_output_tokens": 213
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45063,
            "output_tokens": 45,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45170,
            "output_tokens": 43,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45578,
            "output_tokens": 171,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 46122,
            "output_tokens": 130,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46312,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46402,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46808,
            "output_tokens": 152,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47756,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 47849,
            "output_tokens": 68,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 48282,
            "output_tokens": 295,
            "reasoning_output_tokens": 102
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48512,
            "input_tokens": 48604,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48640,
            "input_tokens": 49549,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49408,
            "input_tokens": 49641,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 50050,
            "output_tokens": 44,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49920,
            "input_tokens": 50263,
            "output_tokens": 130,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50304,
            "input_tokens": 50420,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50304,
            "input_tokens": 51115,
            "output_tokens": 412,
            "reasoning_output_tokens": 232
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51328,
            "input_tokens": 52187,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52096,
            "input_tokens": 52674,
            "output_tokens": 190,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52736,
            "input_tokens": 53466,
            "output_tokens": 167,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53504,
            "input_tokens": 54120,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54016,
            "input_tokens": 54315,
            "output_tokens": 55,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54272,
            "input_tokens": 54539,
            "output_tokens": 924,
            "reasoning_output_tokens": 159
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55296,
            "input_tokens": 55486,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 78,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "bash": 1,
          "exec": 40,
          "devices": 3,
          "device": 5,
          "write": 2,
          "device_stop": 6,
          "edit": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 754 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/handover-block.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/workspace/.kinex/memos/handover-block.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/handover.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-01-20261113/workspace/.kinex/tools/handover.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 150,
        "observed_images": 17
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 303,
        "published_frames": 303,
        "decode_ok": true,
        "duration_s": 7.575,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f20f9bccbf74b1872967f3b1e54a945d3d4c9c3c3aec267fa36652d1fb8b36fa",
        "source_sha256": {
          "third_person": "6fd895d473fefcc9bba79f4b585248befedfabd277dd805c64336c09e7a73e78",
          "left_wrist": "0e958eed535e6f32a9cee276860fb8faf7b737e6a6a28eaf881888ca9bef6116",
          "right_wrist": "2ffb16b79b0842a6228612bbd02483c00de9b511b6a8974eeeaeb80aa5e12c2e"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "400961da599e",
          "task": "robotwin/handover-block-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 303,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 303,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 30160000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 303,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 303,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 30160000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 303,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 303,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 30160000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-02-20261113",
      "variant": "kinex-native",
      "task_key": "task03/02",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "02",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 22.400000000000002,
        "step_limit": 600,
        "steps": 560,
        "success": false,
        "valid": true
      },
      "steps": 560,
      "simulation_time_s": 22.400000000000002,
      "wall_time_s": 1394.884151,
      "timing": {
        "simulation_time_s": 22.400000000000002,
        "wall_time_s": 1394.884151,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "native_instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9620786174463625,
        "cache_read_input_tokens": 868224,
        "cache_reported_input_tokens": 902446,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 902446,
        "cached_input_tokens": 868224,
        "cost_usd": null,
        "fresh_input_tokens": 34222,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 902446,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 868224,
        "known_fresh_input_tokens": 34222,
        "known_input_tokens": 902446,
        "known_output_tokens": 6682,
        "known_reasoning_output_tokens": 2047,
        "output_tokens": 6682,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2047,
        "reasoning_reported_output_tokens": 6682,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 35,
          "cache_write_input_tokens": 35,
          "cache_write_reported_input_tokens": 35,
          "cached_input_tokens": 35,
          "fresh_input_tokens": 35,
          "input_tokens": 35,
          "output_tokens": 35,
          "reasoning_output_tokens": 35,
          "reasoning_reported_output_tokens": 35
        },
        "request_attempts": 35,
        "response_count": 35,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 909128,
        "uncached_input_tokens": 34222,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11820,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11860,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11993,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12041,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12289,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 15327,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15360,
            "input_tokens": 18802,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 19032,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 19224,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 19476,
            "output_tokens": 175,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 20583,
            "output_tokens": 193,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 22786,
            "output_tokens": 344,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23741,
            "output_tokens": 1440,
            "reasoning_output_tokens": 385
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25202,
            "output_tokens": 123,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25833,
            "output_tokens": 154,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26494,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 27033,
            "output_tokens": 149,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27698,
            "output_tokens": 349,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28445,
            "output_tokens": 572,
            "reasoning_output_tokens": 464
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29535,
            "output_tokens": 369,
            "reasoning_output_tokens": 270
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 29964,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30053,
            "output_tokens": 79,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30504,
            "output_tokens": 189,
            "reasoning_output_tokens": 92
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30908,
            "output_tokens": 189,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31124,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31568,
            "output_tokens": 193,
            "reasoning_output_tokens": 139
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32113,
            "output_tokens": 224,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32910,
            "output_tokens": 140,
            "reasoning_output_tokens": 86
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33399,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33863,
            "output_tokens": 104,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34421,
            "output_tokens": 119,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34889,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35410,
            "output_tokens": 596,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36030,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36076,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 43,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 13,
          "bash": 2,
          "exec": 19,
          "devices": 2,
          "device": 1,
          "write": 2,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 560 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/place_dual_shoes.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/workspace/.kinex/memos/place_dual_shoes.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/shoe_motion.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-02-20261113/workspace/.kinex/tools/shoe_motion.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 81,
        "observed_images": 15
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 225,
        "published_frames": 225,
        "decode_ok": true,
        "duration_s": 5.625,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "e6d265f749ed4dc55e7880ff84a1e59c3e7411c8cded01072cd5dd717c74111f",
        "source_sha256": {
          "third_person": "4c3debe6b50a858e7dc38c6a2678036eb4a9473b9be3addaa81b062ea15e2d94",
          "left_wrist": "f2c62f009fee9f74122c1ac8ad6a9309c027972f2a69956e073d0b0963f7f5f7",
          "right_wrist": "8a5b3449ccdc8f1a5fe8842a285ca59c09c41a4baaa5b150d0971845e1ae77ae"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "94d59bfe59c3",
          "task": "robotwin/place-dual-shoes-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 226,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 225,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 22400000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 226,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 225,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 22400000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 226,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 225,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 22400000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-03-20261113",
      "variant": "kinex-native",
      "task_key": "task03/03",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "03",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 2
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 40.36,
        "step_limit": 1700,
        "steps": 1009,
        "success": true,
        "valid": true
      },
      "steps": 1009,
      "simulation_time_s": 40.36,
      "wall_time_s": 2550.836058,
      "timing": {
        "simulation_time_s": 40.36,
        "wall_time_s": 2550.836058,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap in the dustbin the simple black garbage bin.",
      "native_instruction": "Place the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap in the dustbin the simple black garbage bin.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9792914376044775,
        "cache_read_input_tokens": 1972480,
        "cache_reported_input_tokens": 2014191,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2014191,
        "cached_input_tokens": 1972480,
        "cost_usd": null,
        "fresh_input_tokens": 41711,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2014191,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1972480,
        "known_fresh_input_tokens": 41711,
        "known_input_tokens": 2014191,
        "known_output_tokens": 9624,
        "known_reasoning_output_tokens": 2539,
        "output_tokens": 9624,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2539,
        "reasoning_reported_output_tokens": 9624,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 56,
          "cache_write_input_tokens": 56,
          "cache_write_reported_input_tokens": 56,
          "cached_input_tokens": 56,
          "fresh_input_tokens": 56,
          "input_tokens": 56,
          "output_tokens": 56,
          "reasoning_output_tokens": 56,
          "reasoning_reported_output_tokens": 56
        },
        "request_attempts": 56,
        "response_count": 56,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2023815,
        "uncached_input_tokens": 41711,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11842,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11882,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12015,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15688,
            "output_tokens": 182,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 18011,
            "output_tokens": 149,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 21022,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21252,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21443,
            "output_tokens": 266,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 24527,
            "output_tokens": 173,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 26678,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27037,
            "output_tokens": 445,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 28079,
            "output_tokens": 1246,
            "reasoning_output_tokens": 349
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29348,
            "output_tokens": 190,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29822,
            "output_tokens": 194,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30701,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30782,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31597,
            "output_tokens": 54,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32399,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32527,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33118,
            "output_tokens": 111,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33258,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33969,
            "output_tokens": 60,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34724,
            "output_tokens": 257,
            "reasoning_output_tokens": 122
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35718,
            "output_tokens": 282,
            "reasoning_output_tokens": 169
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36060,
            "output_tokens": 47,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36169,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36586,
            "output_tokens": 195,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36810,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37295,
            "output_tokens": 72,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37732,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38204,
            "output_tokens": 75,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38448,
            "output_tokens": 121,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38991,
            "output_tokens": 313,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39668,
            "output_tokens": 163,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40222,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40646,
            "output_tokens": 436,
            "reasoning_output_tokens": 289
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41585,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42006,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42474,
            "output_tokens": 182,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 43101,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 43527,
            "output_tokens": 364,
            "reasoning_output_tokens": 246
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 44055,
            "output_tokens": 160,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44693,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 45113,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45579,
            "output_tokens": 263,
            "reasoning_output_tokens": 163
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 46319,
            "output_tokens": 95,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46776,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 47324,
            "output_tokens": 280,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47488,
            "input_tokens": 47801,
            "output_tokens": 225,
            "reasoning_output_tokens": 106
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47872,
            "input_tokens": 48500,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 48921,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49390,
            "output_tokens": 169,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49408,
            "input_tokens": 50002,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49920,
            "input_tokens": 50205,
            "output_tokens": 91,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50176,
            "input_tokens": 50680,
            "output_tokens": 1163,
            "reasoning_output_tokens": 186
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51712,
            "input_tokens": 51870,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 69,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 17,
          "find": 1,
          "devices": 5,
          "device": 2,
          "exec": 34,
          "write": 2,
          "edit": 4,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 1009 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/manipulate.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-03-20261113/workspace/.kinex/tools/manipulate.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 128,
        "observed_images": 28
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 1700,
        "max_control_steps": 1700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 405,
        "published_frames": 405,
        "decode_ok": true,
        "duration_s": 10.125,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "45a8af1166e4657389c6e7c7fee9542cc234195789c0e271493a9290b55e5a99",
        "source_sha256": {
          "third_person": "d74468628ef0c3a64cc34922dc61ae6713b2ee15949fd0eb953ea10d95ac4944",
          "left_wrist": "bd1004895f6b8a055f7cab90d613dc78d76594dccefca2339c2e935ef93cc3e3",
          "right_wrist": "88c8e69238ecd5887876a47028ae3262cdce53b63ad4d751f150741ff9b868a2"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ca1b0ad6e704",
          "task": "robotwin/put-bottles-dustbin-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 405,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 405,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 40360000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 405,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 405,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 40360000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 405,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 405,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 40360000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-04-20261113",
      "variant": "kinex-native",
      "task_key": "task03/04",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "04",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.24,
        "step_limit": 500,
        "steps": 381,
        "success": false,
        "valid": true
      },
      "steps": 381,
      "simulation_time_s": 15.24,
      "wall_time_s": 1774.253756,
      "timing": {
        "simulation_time_s": 15.24,
        "wall_time_s": 1774.253756,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "native_instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9588585537854595,
        "cache_read_input_tokens": 1683840,
        "cache_reported_input_tokens": 1756088,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1756088,
        "cached_input_tokens": 1683840,
        "cost_usd": null,
        "fresh_input_tokens": 72248,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1756088,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1683840,
        "known_fresh_input_tokens": 72248,
        "known_input_tokens": 1756088,
        "known_output_tokens": 11059,
        "known_reasoning_output_tokens": 4597,
        "output_tokens": 11059,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 4597,
        "reasoning_reported_output_tokens": 11059,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 55,
          "cache_write_input_tokens": 55,
          "cache_write_reported_input_tokens": 55,
          "cached_input_tokens": 55,
          "fresh_input_tokens": 55,
          "input_tokens": 55,
          "output_tokens": 55,
          "reasoning_output_tokens": 55,
          "reasoning_reported_output_tokens": 55
        },
        "request_attempts": 55,
        "response_count": 55,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1767147,
        "uncached_input_tokens": 72248,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11827,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11867,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12000,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12050,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12298,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12449,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 16975,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 18868,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19098,
            "output_tokens": 160,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 21252,
            "output_tokens": 106,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21568,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 21976,
            "output_tokens": 341,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22852,
            "output_tokens": 327,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23782,
            "output_tokens": 144,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24392,
            "output_tokens": 89,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 25660,
            "output_tokens": 1060,
            "reasoning_output_tokens": 140
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26743,
            "output_tokens": 104,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27125,
            "output_tokens": 148,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27333,
            "output_tokens": 43,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27438,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28030,
            "output_tokens": 185,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28244,
            "output_tokens": 119,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28871,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28960,
            "output_tokens": 110,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29099,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29715,
            "output_tokens": 183,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30552,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31210,
            "output_tokens": 163,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 32021,
            "output_tokens": 195,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 33756,
            "output_tokens": 175,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34274,
            "output_tokens": 555,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34892,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35255,
            "output_tokens": 423,
            "reasoning_output_tokens": 286
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 36365,
            "output_tokens": 138,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37367,
            "output_tokens": 341,
            "reasoning_output_tokens": 231
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38350,
            "output_tokens": 47,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38459,
            "output_tokens": 536,
            "reasoning_output_tokens": 411
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39690,
            "output_tokens": 333,
            "reasoning_output_tokens": 240
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40815,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 40906,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41324,
            "output_tokens": 666,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42095,
            "output_tokens": 149,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42516,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 43219,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43412,
            "output_tokens": 364,
            "reasoning_output_tokens": 224
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44299,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44491,
            "output_tokens": 113,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44928,
            "output_tokens": 664,
            "reasoning_output_tokens": 305
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45854,
            "output_tokens": 663,
            "reasoning_output_tokens": 608
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 47008,
            "output_tokens": 427,
            "reasoning_output_tokens": 319
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47966,
            "output_tokens": 62,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47872,
            "input_tokens": 48090,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48000,
            "input_tokens": 48451,
            "output_tokens": 245,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48640,
            "input_tokens": 48725,
            "output_tokens": 577,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49152,
            "input_tokens": 49326,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 63,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "bash": 1,
          "exec": 24,
          "devices": 7,
          "device": 3,
          "write": 2,
          "device_stop": 5,
          "edit": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 381 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/dual_arm_scanning.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/workspace/.kinex/memos/dual_arm_scanning.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_manipulate.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-04-20261113/workspace/.kinex/tools/dual_manipulate.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 121,
        "observed_images": 19
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 154,
        "published_frames": 154,
        "decode_ok": true,
        "duration_s": 3.85,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c3cd09bf8d16b3334438faedf6c93a840be7342cd779ebb255f340ffaffcccc5",
        "source_sha256": {
          "third_person": "5181e63828829c55ea7a439e2add6c8046c5099a367e2513fcca9db7886f26a2",
          "left_wrist": "b9bea352793456744478fd9a94c918442321e1ce0a61520ec06d8daebd5adc01",
          "right_wrist": "7aecffb4cffe77d74f9ee2550b0048503049c3824392709ed6a81ff91cadacd0"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "dd2508359c14",
          "task": "robotwin/scan-object-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 154,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 154,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15240000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 154,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 154,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15240000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 154,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 154,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15240000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-05-20261113",
      "variant": "kinex-native",
      "task_key": "task03/05",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "05",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 32.36,
        "step_limit": 1200,
        "steps": 809,
        "success": true,
        "valid": true
      },
      "steps": 809,
      "simulation_time_s": 32.36,
      "wall_time_s": 2668.778956,
      "timing": {
        "simulation_time_s": 32.36,
        "wall_time_s": 2668.778956,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "native_instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.976225124914745,
        "cache_read_input_tokens": 2002432,
        "cache_reported_input_tokens": 2051199,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2051199,
        "cached_input_tokens": 2002432,
        "cost_usd": null,
        "fresh_input_tokens": 48767,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2051199,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2002432,
        "known_fresh_input_tokens": 48767,
        "known_input_tokens": 2051199,
        "known_output_tokens": 9278,
        "known_reasoning_output_tokens": 2050,
        "output_tokens": 9278,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2050,
        "reasoning_reported_output_tokens": 9278,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 56,
          "cache_write_input_tokens": 56,
          "cache_write_reported_input_tokens": 56,
          "cached_input_tokens": 56,
          "fresh_input_tokens": 56,
          "input_tokens": 56,
          "output_tokens": 56,
          "reasoning_output_tokens": 56,
          "reasoning_reported_output_tokens": 56
        },
        "request_attempts": 56,
        "response_count": 56,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2060477,
        "uncached_input_tokens": 48767,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11821,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11861,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11994,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 16921,
            "output_tokens": 217,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17024,
            "input_tokens": 20048,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 22579,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22809,
            "output_tokens": 167,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 25665,
            "output_tokens": 177,
            "reasoning_output_tokens": 133
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26052,
            "output_tokens": 187,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 27691,
            "output_tokens": 59,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28119,
            "output_tokens": 372,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 29175,
            "output_tokens": 592,
            "reasoning_output_tokens": 185
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29983,
            "output_tokens": 185,
            "reasoning_output_tokens": 143
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 31347,
            "output_tokens": 1185,
            "reasoning_output_tokens": 116
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32552,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33297,
            "output_tokens": 88,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33411,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33799,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34193,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34540,
            "output_tokens": 126,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34973,
            "output_tokens": 691,
            "reasoning_output_tokens": 157
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35685,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36064,
            "output_tokens": 133,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36368,
            "output_tokens": 196,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36624,
            "output_tokens": 61,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 36747,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37167,
            "output_tokens": 124,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37468,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37873,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38225,
            "output_tokens": 249,
            "reasoning_output_tokens": 150
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 39282,
            "output_tokens": 49,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39393,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39506,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39848,
            "output_tokens": 107,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40127,
            "output_tokens": 238,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 41073,
            "output_tokens": 72,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41893,
            "output_tokens": 120,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42039,
            "output_tokens": 171,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 43107,
            "output_tokens": 464,
            "reasoning_output_tokens": 325
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 44371,
            "output_tokens": 43,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44476,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44821,
            "output_tokens": 98,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 44945,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45301,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45702,
            "output_tokens": 183,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46193,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46549,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 46949,
            "output_tokens": 124,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47356,
            "output_tokens": 169,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47360,
            "input_tokens": 47838,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 48192,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48128,
            "input_tokens": 48591,
            "output_tokens": 231,
            "reasoning_output_tokens": 96
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48640,
            "input_tokens": 49128,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49323,
            "output_tokens": 91,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49280,
            "input_tokens": 49674,
            "output_tokens": 773,
            "reasoning_output_tokens": 118
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50432,
            "input_tokens": 50471,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 66,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 14,
          "find": 1,
          "exec": 33,
          "devices": 3,
          "device": 1,
          "write": 3,
          "edit": 4,
          "device_stop": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 809 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack-bowls.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/workspace/.kinex/memos/stack-bowls.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/bowl_geometry.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/workspace/.kinex/tools/bowl_geometry.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tabletop.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-05-20261113/workspace/.kinex/tools/tabletop.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 125,
        "observed_images": 20
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 325,
        "published_frames": 325,
        "decode_ok": true,
        "duration_s": 8.125,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "8f5d75a28a46e42959375f53b43567601adbb2aabd26fb41a53913e16ea911b0",
        "source_sha256": {
          "third_person": "a5dc796f24f4e5192be42ee44b9be4b4174f62ef5b0d00b928f389ef52c60572",
          "left_wrist": "afe2fbeddfb8683d247b27c679727d21eefe34ef0e736d3581c4c1f9a2111c8a",
          "right_wrist": "fa80a0b6d202faf1e91185cde49c61a96ad26d1fd044ba7f051522c9e533be84"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "9ef80e5226f7",
          "task": "robotwin/stack-bowls-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 325,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 325,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 32360000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 325,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 325,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 32360000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 325,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 325,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 32360000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-06-20261113",
      "variant": "kinex-native",
      "task_key": "task03/06",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "06",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 33.28,
        "step_limit": 1200,
        "steps": 832,
        "success": true,
        "valid": true
      },
      "steps": 832,
      "simulation_time_s": 33.28,
      "wall_time_s": 3381.668175,
      "timing": {
        "simulation_time_s": 33.28,
        "wall_time_s": 3381.668175,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "native_instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9828308976376987,
        "cache_read_input_tokens": 2243456,
        "cache_reported_input_tokens": 2282647,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2282647,
        "cached_input_tokens": 2243456,
        "cost_usd": null,
        "fresh_input_tokens": 39191,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2282647,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2243456,
        "known_fresh_input_tokens": 39191,
        "known_input_tokens": 2282647,
        "known_output_tokens": 9936,
        "known_reasoning_output_tokens": 2302,
        "output_tokens": 9936,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2302,
        "reasoning_reported_output_tokens": 9936,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 67,
          "cache_write_input_tokens": 67,
          "cache_write_reported_input_tokens": 67,
          "cached_input_tokens": 67,
          "fresh_input_tokens": 67,
          "input_tokens": 67,
          "output_tokens": 67,
          "reasoning_output_tokens": 67,
          "reasoning_reported_output_tokens": 67
        },
        "request_attempts": 67,
        "response_count": 67,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2292583,
        "uncached_input_tokens": 39191,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11821,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11861,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11994,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12044,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13316,
            "output_tokens": 205,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13440,
            "input_tokens": 17177,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17152,
            "input_tokens": 18439,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 19009,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 19239,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 19432,
            "output_tokens": 125,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 20899,
            "output_tokens": 272,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21706,
            "output_tokens": 174,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22949,
            "output_tokens": 517,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 24405,
            "output_tokens": 188,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 27698,
            "output_tokens": 1797,
            "reasoning_output_tokens": 397
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29518,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29905,
            "output_tokens": 158,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30123,
            "output_tokens": 44,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30229,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30647,
            "output_tokens": 293,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30969,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31275,
            "output_tokens": 54,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31401,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31723,
            "output_tokens": 85,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31959,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32269,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32820,
            "output_tokens": 342,
            "reasoning_output_tokens": 267
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33404,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33718,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33923,
            "output_tokens": 128,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34298,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 35173,
            "output_tokens": 50,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35245,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35795,
            "output_tokens": 136,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36174,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36490,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36696,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37016,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37891,
            "output_tokens": 60,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 37973,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38066,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38615,
            "output_tokens": 168,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39025,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39340,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39545,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40369,
            "output_tokens": 237,
            "reasoning_output_tokens": 165
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40850,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41172,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41722,
            "output_tokens": 113,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 41853,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42244,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42582,
            "output_tokens": 333,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43206,
            "output_tokens": 130,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43577,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 43787,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 44110,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44659,
            "output_tokens": 46,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 44874,
            "output_tokens": 294,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45314,
            "output_tokens": 289,
            "reasoning_output_tokens": 207
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45846,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 46400,
            "output_tokens": 339,
            "reasoning_output_tokens": 257
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 46982,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47183,
            "output_tokens": 81,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 47326,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47877,
            "output_tokens": 501,
            "reasoning_output_tokens": 234
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48128,
            "input_tokens": 48407,
            "output_tokens": 662,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49093,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 76,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 15,
          "find": 1,
          "exec": 43,
          "devices": 2,
          "device": 4,
          "write": 2,
          "device_stop": 4,
          "edit": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 832 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack_blocks.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/workspace/.kinex/memos/stack_blocks.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/stack_blocks.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-06-20261113/workspace/.kinex/tools/stack_blocks.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 146,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 334,
        "published_frames": 334,
        "decode_ok": true,
        "duration_s": 8.35,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "9e6fea6a093af1888bdf4ec7b1d8bef79e0be4acd58a10c81ba0373932f37891",
        "source_sha256": {
          "third_person": "af7f6a37ef2eef80e08b23ddfc2bf98ed0516e9add35f4d69889a7f875e75080",
          "left_wrist": "f9c3ca90a144d6e9550b44a481cdb99fc33313fe20685a6fa8fbe054b03e0ea3",
          "right_wrist": "f1e61a058474368ec31e41f064023273d78e1e7b299c8c0a36ff601969cd86d1"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "a4b1845cfaa2",
          "task": "robotwin/stack-blocks-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 334,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 334,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33280000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 334,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 334,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33280000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 334,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 334,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33280000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-07-20261113",
      "variant": "kinex-native",
      "task_key": "task03/07",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "07",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 35.96,
        "step_limit": 900,
        "steps": 899,
        "success": false,
        "valid": true
      },
      "steps": 899,
      "simulation_time_s": 35.96,
      "wall_time_s": 2464.009277,
      "timing": {
        "simulation_time_s": 35.96,
        "wall_time_s": 2464.009277,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the cylindrical mug with handle on the table, rotate it, set it down in the middle, then hang it onto the rack with straight and slanted rods.",
      "native_instruction": "Grab the cylindrical mug with handle on the table, rotate it, set it down in the middle, then hang it onto the rack with straight and slanted rods.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.98595451604346,
        "cache_read_input_tokens": 3512320,
        "cache_reported_input_tokens": 3562355,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 3562355,
        "cached_input_tokens": 3512320,
        "cost_usd": null,
        "fresh_input_tokens": 50035,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 3562355,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 3512320,
        "known_fresh_input_tokens": 50035,
        "known_input_tokens": 3562355,
        "known_output_tokens": 16521,
        "known_reasoning_output_tokens": 5838,
        "output_tokens": 16521,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 5838,
        "reasoning_reported_output_tokens": 16521,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 88,
          "cache_write_input_tokens": 88,
          "cache_write_reported_input_tokens": 88,
          "cached_input_tokens": 88,
          "fresh_input_tokens": 88,
          "input_tokens": 88,
          "output_tokens": 88,
          "reasoning_output_tokens": 88,
          "reasoning_reported_output_tokens": 88
        },
        "request_attempts": 88,
        "response_count": 88,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 3578876,
        "uncached_input_tokens": 50035,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11835,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11968,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12016,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12264,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12415,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16941,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 18204,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18774,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 19004,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 19196,
            "output_tokens": 151,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 21559,
            "output_tokens": 82,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 22010,
            "output_tokens": 371,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22916,
            "output_tokens": 363,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 24472,
            "output_tokens": 153,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25804,
            "output_tokens": 373,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 27452,
            "output_tokens": 1378,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28851,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29135,
            "output_tokens": 63,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29893,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30079,
            "output_tokens": 47,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30188,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30605,
            "output_tokens": 178,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30810,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31321,
            "output_tokens": 153,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 32222,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32312,
            "output_tokens": 224,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32563,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32824,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33195,
            "output_tokens": 125,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33607,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33800,
            "output_tokens": 250,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34710,
            "output_tokens": 63,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35585,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35714,
            "output_tokens": 177,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36593,
            "output_tokens": 218,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36838,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37231,
            "output_tokens": 195,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37726,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38098,
            "output_tokens": 130,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38521,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38713,
            "output_tokens": 574,
            "reasoning_output_tokens": 479
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39567,
            "output_tokens": 259,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40741,
            "output_tokens": 273,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41446,
            "output_tokens": 394,
            "reasoning_output_tokens": 250
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 42120,
            "output_tokens": 71,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42213,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42407,
            "output_tokens": 92,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42561,
            "output_tokens": 236,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 42951,
            "output_tokens": 466,
            "reasoning_output_tokens": 169
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43965,
            "output_tokens": 191,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44183,
            "output_tokens": 187,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44521,
            "output_tokens": 156,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44959,
            "output_tokens": 175,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45156,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45347,
            "output_tokens": 287,
            "reasoning_output_tokens": 155
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45914,
            "output_tokens": 242,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45952,
            "input_tokens": 46500,
            "output_tokens": 185,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 46965,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47682,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47488,
            "input_tokens": 47874,
            "output_tokens": 375,
            "reasoning_output_tokens": 243
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48128,
            "input_tokens": 48530,
            "output_tokens": 163,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48512,
            "input_tokens": 49249,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49152,
            "input_tokens": 49440,
            "output_tokens": 119,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 49622,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 49814,
            "output_tokens": 120,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49792,
            "input_tokens": 50102,
            "output_tokens": 312,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50304,
            "input_tokens": 52094,
            "output_tokens": 329,
            "reasoning_output_tokens": 228
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52352,
            "input_tokens": 52702,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52736,
            "input_tokens": 53422,
            "output_tokens": 46,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53376,
            "input_tokens": 53638,
            "output_tokens": 898,
            "reasoning_output_tokens": 770
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54400,
            "input_tokens": 54813,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54784,
            "input_tokens": 55007,
            "output_tokens": 377,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55296,
            "input_tokens": 56894,
            "output_tokens": 478,
            "reasoning_output_tokens": 385
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57216,
            "input_tokens": 57653,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57600,
            "input_tokens": 57845,
            "output_tokens": 152,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57856,
            "input_tokens": 58277,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58240,
            "input_tokens": 58471,
            "output_tokens": 205,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58496,
            "input_tokens": 59219,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59136,
            "input_tokens": 59412,
            "output_tokens": 474,
            "reasoning_output_tokens": 337
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59776,
            "input_tokens": 60170,
            "output_tokens": 159,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60160,
            "input_tokens": 60874,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60800,
            "input_tokens": 61069,
            "output_tokens": 849,
            "reasoning_output_tokens": 715
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61696,
            "input_tokens": 62198,
            "output_tokens": 159,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62208,
            "input_tokens": 62903,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62720,
            "input_tokens": 63095,
            "output_tokens": 82,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62976,
            "input_tokens": 63228,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 63104,
            "input_tokens": 63420,
            "output_tokens": 715,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 64128,
            "input_tokens": 64158,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 95,
        "model_tool_calls_by_name": {
          "read": 17,
          "ls": 1,
          "bash": 1,
          "exec": 42,
          "devices": 4,
          "device": 18,
          "write": 2,
          "device_stop": 5,
          "edit": 5
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 899 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/hanging-mug.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/workspace/.kinex/memos/hanging-mug.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_motion.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-07-20261113/workspace/.kinex/tools/cartesian_motion.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 186,
        "observed_images": 24
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 361,
        "published_frames": 361,
        "decode_ok": true,
        "duration_s": 9.025,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "7873e3ea4f3eb7be61f996fface8129da169cee62ba47081d1c8dfbb6d90347d",
        "source_sha256": {
          "third_person": "c043ed6ad4abf622b144931a11bea40bd44d7980c34fdece5a77c737ffb0c8c1",
          "left_wrist": "fafaab794fad0bb3ef0be18d354412f2e73ca95feec8e855a68940bf76f8628a",
          "right_wrist": "80d4f7c54502bd308f581f6275df3721116a6fafbd3ec88739e53aeb4c5ea15b"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "5dd8e81bba25",
          "task": "robotwin/hanging-mug-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 361,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 361,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 35960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 361,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 361,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 35960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 361,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 361,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 35960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-08-20261113",
      "variant": "kinex-native",
      "task_key": "task03/08",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "08",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 27.240000000000002,
        "step_limit": 700,
        "steps": 681,
        "success": false,
        "valid": true
      },
      "steps": 681,
      "simulation_time_s": 27.240000000000002,
      "wall_time_s": 3097.082165,
      "timing": {
        "simulation_time_s": 27.240000000000002,
        "wall_time_s": 3097.082165,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Open the cabinet with smooth finish's drawer and set the blue soap with printed patterns inside with the other arm.",
      "native_instruction": "Open the cabinet with smooth finish's drawer and set the blue soap with printed patterns inside with the other arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9850638403695363,
        "cache_read_input_tokens": 2867840,
        "cache_reported_input_tokens": 2911324,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2911324,
        "cached_input_tokens": 2867840,
        "cost_usd": null,
        "fresh_input_tokens": 43484,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2911324,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2867840,
        "known_fresh_input_tokens": 43484,
        "known_input_tokens": 2911324,
        "known_output_tokens": 11488,
        "known_reasoning_output_tokens": 4552,
        "output_tokens": 11488,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 4552,
        "reasoning_reported_output_tokens": 11488,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 88,
          "cache_write_input_tokens": 88,
          "cache_write_reported_input_tokens": 88,
          "cached_input_tokens": 88,
          "fresh_input_tokens": 88,
          "input_tokens": 88,
          "output_tokens": 88,
          "reasoning_output_tokens": 88,
          "reasoning_reported_output_tokens": 88
        },
        "request_attempts": 88,
        "response_count": 88,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2922812,
        "uncached_input_tokens": 43484,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11827,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11958,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12006,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12254,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12405,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 12868,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12800,
            "input_tokens": 13682,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13568,
            "input_tokens": 14499,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14336,
            "input_tokens": 15325,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15232,
            "input_tokens": 16153,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 16945,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 18208,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 18438,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 18628,
            "output_tokens": 84,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 19243,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 19495,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 20299,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 21104,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21592,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 21952,
            "output_tokens": 118,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22959,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23367,
            "output_tokens": 233,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23620,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24179,
            "output_tokens": 91,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24736,
            "output_tokens": 710,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25466,
            "output_tokens": 139,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 26223,
            "output_tokens": 85,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26334,
            "output_tokens": 139,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26871,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27232,
            "output_tokens": 106,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27755,
            "output_tokens": 159,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28286,
            "output_tokens": 64,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28826,
            "output_tokens": 190,
            "reasoning_output_tokens": 143
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29128,
            "output_tokens": 119,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29307,
            "output_tokens": 39,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29408,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29986,
            "output_tokens": 210,
            "reasoning_output_tokens": 120
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30881,
            "output_tokens": 193,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31307,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31500,
            "output_tokens": 213,
            "reasoning_output_tokens": 100
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32612,
            "output_tokens": 147,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33196,
            "output_tokens": 101,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33404,
            "output_tokens": 170,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 34354,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34443,
            "output_tokens": 137,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34606,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35008,
            "output_tokens": 99,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35277,
            "output_tokens": 393,
            "reasoning_output_tokens": 259
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36035,
            "output_tokens": 110,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36375,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36733,
            "output_tokens": 151,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37089,
            "output_tokens": 61,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37172,
            "output_tokens": 52,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37919,
            "output_tokens": 207,
            "reasoning_output_tokens": 143
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38152,
            "output_tokens": 105,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38395,
            "output_tokens": 138,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38779,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 38971,
            "output_tokens": 98,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39830,
            "output_tokens": 80,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40480,
            "output_tokens": 275,
            "reasoning_output_tokens": 164
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41449,
            "output_tokens": 132,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41812,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42005,
            "output_tokens": 308,
            "reasoning_output_tokens": 255
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42420,
            "output_tokens": 215,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 43005,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43199,
            "output_tokens": 446,
            "reasoning_output_tokens": 295
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 44012,
            "output_tokens": 110,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44144,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44337,
            "output_tokens": 250,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44835,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45028,
            "output_tokens": 305,
            "reasoning_output_tokens": 146
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45704,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 45898,
            "output_tokens": 533,
            "reasoning_output_tokens": 414
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46667,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 46859,
            "output_tokens": 361,
            "reasoning_output_tokens": 245
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 47450,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47360,
            "input_tokens": 47643,
            "output_tokens": 337,
            "reasoning_output_tokens": 225
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47872,
            "input_tokens": 48207,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48128,
            "input_tokens": 48398,
            "output_tokens": 236,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48512,
            "input_tokens": 48872,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48768,
            "input_tokens": 49066,
            "output_tokens": 591,
            "reasoning_output_tokens": 428
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 49888,
            "output_tokens": 76,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49792,
            "input_tokens": 50071,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49920,
            "input_tokens": 50265,
            "output_tokens": 367,
            "reasoning_output_tokens": 270
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50432,
            "input_tokens": 50868,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50816,
            "input_tokens": 51061,
            "output_tokens": 121,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51072,
            "input_tokens": 51244,
            "output_tokens": 569,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51712,
            "input_tokens": 51835,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 87,
        "model_tool_calls_by_name": {
          "read": 15,
          "ls": 1,
          "bash": 1,
          "exec": 36,
          "devices": 8,
          "device": 15,
          "write": 3,
          "edit": 3,
          "device_stop": 5
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 681 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/workspace/.kinex/memos/robotwin.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/workspace/.kinex/tools/move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/observe.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-08-20261113/workspace/.kinex/tools/observe.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 178,
        "observed_images": 33
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 274,
        "published_frames": 274,
        "decode_ok": true,
        "duration_s": 6.85,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "7b99654a9039baaf5c5826fc1b248507fbb263dc63d7eb95e50d20bcde61335b",
        "source_sha256": {
          "third_person": "fb5d70a3d8f450a69a5235041087113893de5c4e4be58058bf43e5162d7e29e8",
          "left_wrist": "5458b6357a5393a1e1eac07ae145b5618dd5de223f17f14252463a8acbeb5b48",
          "right_wrist": "94aefe7aad1bb2b8d55bc4951b56011c74b9bf75bb297b0de942a481fd8cad0e"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "1239428b5df0",
          "task": "robotwin/put-object-cabinet-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 274,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 274,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27240000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 274,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 274,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27240000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 274,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 274,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27240000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-09-20261113",
      "variant": "kinex-native",
      "task_key": "task03/09",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "09",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 30.12,
        "step_limit": 1200,
        "steps": 753,
        "success": true,
        "valid": true
      },
      "steps": 753,
      "simulation_time_s": 30.12,
      "wall_time_s": 2012.939236,
      "timing": {
        "simulation_time_s": 30.12,
        "wall_time_s": 2012.939236,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move large block, medium block, and small block to the center table.",
      "native_instruction": "Move large block, medium block, and small block to the center table.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9687074401752943,
        "cache_read_input_tokens": 1231232,
        "cache_reported_input_tokens": 1271005,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1271005,
        "cached_input_tokens": 1231232,
        "cost_usd": null,
        "fresh_input_tokens": 39773,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1271005,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1231232,
        "known_fresh_input_tokens": 39773,
        "known_input_tokens": 1271005,
        "known_output_tokens": 7279,
        "known_reasoning_output_tokens": 1241,
        "output_tokens": 7279,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1241,
        "reasoning_reported_output_tokens": 7279,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 43,
          "cache_write_input_tokens": 43,
          "cache_write_reported_input_tokens": 43,
          "cached_input_tokens": 43,
          "fresh_input_tokens": 43,
          "input_tokens": 43,
          "output_tokens": 43,
          "reasoning_output_tokens": 43,
          "reasoning_reported_output_tokens": 43
        },
        "request_attempts": 43,
        "response_count": 43,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1278284,
        "uncached_input_tokens": 39773,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11821,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11861,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11994,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12042,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14773,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 17811,
            "output_tokens": 136,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 20513,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 21776,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22006,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22196,
            "output_tokens": 213,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23450,
            "output_tokens": 339,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24324,
            "output_tokens": 308,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25854,
            "output_tokens": 122,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 27155,
            "output_tokens": 903,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28079,
            "output_tokens": 371,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28791,
            "output_tokens": 248,
            "reasoning_output_tokens": 172
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29360,
            "output_tokens": 118,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29841,
            "output_tokens": 92,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30258,
            "output_tokens": 171,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30489,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30544,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 31031,
            "output_tokens": 127,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31506,
            "output_tokens": 210,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32399,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32456,
            "output_tokens": 179,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32938,
            "output_tokens": 146,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33487,
            "output_tokens": 172,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34440,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34534,
            "output_tokens": 274,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34835,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35321,
            "output_tokens": 150,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35904,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36307,
            "output_tokens": 115,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 37107,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37165,
            "output_tokens": 237,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37904,
            "output_tokens": 161,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38459,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38908,
            "output_tokens": 100,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39332,
            "output_tokens": 214,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39688,
            "output_tokens": 69,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40017,
            "output_tokens": 609,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40845,
            "output_tokens": 617,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41484,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 51,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 14,
          "find": 1,
          "exec": 22,
          "devices": 4,
          "device": 1,
          "write": 2,
          "device_stop": 4,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 753 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/blocks-ranking.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/workspace/.kinex/memos/blocks-ranking.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/blocks_motion.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-09-20261113/workspace/.kinex/tools/blocks_motion.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 97,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 302,
        "published_frames": 302,
        "decode_ok": true,
        "duration_s": 7.55,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "d1652052435843e7bb20c96b2ba084091f135f33af817c73868a49df61b9b895",
        "source_sha256": {
          "third_person": "9c78e0871b9d30c52b6d65429c108c36454f32a305ec28bb244a90ec15d086cb",
          "left_wrist": "f01407a04b5faae466b4a8a819df6e417977a4e46b3892c273e88f176ed8bb82",
          "right_wrist": "efeb8adf0f812b9bd59ea58f65ee4e3ab08eff4349d492a262df40cf415efc27"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "3852e7ab699f",
          "task": "robotwin/blocks-ranking-size-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 303,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 302,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 30120000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 303,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 302,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 30120000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 303,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 302,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 30120000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-10-20261113",
      "variant": "kinex-native",
      "task_key": "task03/10",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "10",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.68,
        "step_limit": 400,
        "steps": 242,
        "success": true,
        "valid": true
      },
      "steps": 242,
      "simulation_time_s": 9.68,
      "wall_time_s": 1165.000625,
      "timing": {
        "simulation_time_s": 9.68,
        "wall_time_s": 1165.000625,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Hold the dark gray cooking kitchenpot firmly and lift it",
      "native_instruction": "Hold the dark gray cooking kitchenpot firmly and lift it",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9348514526909852,
        "cache_read_input_tokens": 1032320,
        "cache_reported_input_tokens": 1104261,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1104261,
        "cached_input_tokens": 1032320,
        "cost_usd": null,
        "fresh_input_tokens": 71941,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1104261,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1032320,
        "known_fresh_input_tokens": 71941,
        "known_input_tokens": 1104261,
        "known_output_tokens": 6139,
        "known_reasoning_output_tokens": 1022,
        "output_tokens": 6139,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1022,
        "reasoning_reported_output_tokens": 6139,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 38,
          "cache_write_input_tokens": 38,
          "cache_write_reported_input_tokens": 38,
          "cached_input_tokens": 38,
          "fresh_input_tokens": 38,
          "input_tokens": 38,
          "output_tokens": 38,
          "reasoning_output_tokens": 38,
          "reasoning_reported_output_tokens": 38
        },
        "request_attempts": 38,
        "response_count": 38,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1110400,
        "uncached_input_tokens": 71941,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11814,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11854,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11987,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14009,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13952,
            "input_tokens": 14196,
            "output_tokens": 223,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14336,
            "input_tokens": 18867,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 20758,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 20988,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21181,
            "output_tokens": 122,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 22991,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23243,
            "output_tokens": 60,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23672,
            "output_tokens": 303,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24657,
            "output_tokens": 331,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 26640,
            "output_tokens": 421,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 30210,
            "output_tokens": 977,
            "reasoning_output_tokens": 249
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31209,
            "output_tokens": 144,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31459,
            "output_tokens": 130,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31802,
            "output_tokens": 208,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 32070,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32159,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32754,
            "output_tokens": 206,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 32988,
            "output_tokens": 146,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33778,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33868,
            "output_tokens": 66,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 33962,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34336,
            "output_tokens": 298,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 35055,
            "output_tokens": 56,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35173,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35364,
            "output_tokens": 375,
            "reasoning_output_tokens": 240
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 36133,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36569,
            "output_tokens": 150,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37113,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37538,
            "output_tokens": 159,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37839,
            "output_tokens": 86,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38246,
            "output_tokens": 281,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38902,
            "output_tokens": 247,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39177,
            "output_tokens": 501,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39700,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 45,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 13,
          "find": 1,
          "bash": 1,
          "exec": 16,
          "devices": 3,
          "device": 2,
          "write": 2,
          "device_stop": 3,
          "edit": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 242 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/lift-pot.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/workspace/.kinex/memos/lift-pot.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_cartesian.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-10-20261113/workspace/.kinex/tools/dual_cartesian.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 86,
        "observed_images": 4
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 98,
        "published_frames": 98,
        "decode_ok": true,
        "duration_s": 2.45,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "aba8ce58176a05ab653a36d980060156a37f7e7b56a8b60d7e32dd7d42eaad86",
        "source_sha256": {
          "third_person": "8020b68b6a1c67c03237a317b1b49195a63a857ca8de390110d4a6e6d19e8680",
          "left_wrist": "7af9b95bad314d954e387ee12c63d06e96ec4393fe5248f325be91596706fb75",
          "right_wrist": "49bc77d5e2c727891e0cab3e84c1f870341b69a448a4a119f1a6a47ae0b8c33c"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "2180946df853",
          "task": "robotwin/lift-pot-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 98,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9680000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 98,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9680000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 98,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9680000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-11-20261113",
      "variant": "kinex-native",
      "task_key": "task03/11",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "11",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 2
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 9.120000000000001,
        "step_limit": 400,
        "steps": 228,
        "success": false,
        "valid": true
      },
      "steps": 228,
      "simulation_time_s": 9.120000000000001,
      "wall_time_s": 581.756195,
      "timing": {
        "simulation_time_s": 9.120000000000001,
        "wall_time_s": 581.756195,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the green plastic bottle with ridged bottom carefully using the right arm",
      "native_instruction": "Pick up the green plastic bottle with ridged bottom carefully using the right arm",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9127742386646996,
        "cache_read_input_tokens": 595200,
        "cache_reported_input_tokens": 652078,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 652078,
        "cached_input_tokens": 595200,
        "cost_usd": null,
        "fresh_input_tokens": 56878,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 652078,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 595200,
        "known_fresh_input_tokens": 56878,
        "known_input_tokens": 652078,
        "known_output_tokens": 5662,
        "known_reasoning_output_tokens": 920,
        "output_tokens": 5662,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 920,
        "reasoning_reported_output_tokens": 5662,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 32,
          "cache_write_input_tokens": 32,
          "cache_write_reported_input_tokens": 32,
          "cached_input_tokens": 32,
          "fresh_input_tokens": 32,
          "input_tokens": 32,
          "output_tokens": 32,
          "reasoning_output_tokens": 32,
          "reasoning_reported_output_tokens": 32
        },
        "request_attempts": 32,
        "response_count": 32,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 657740,
        "uncached_input_tokens": 56878,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11818,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11951,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11999,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 12247,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 13064,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12928,
            "input_tokens": 13527,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13440,
            "input_tokens": 14355,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14208,
            "input_tokens": 15147,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14976,
            "input_tokens": 15298,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15232,
            "input_tokens": 16560,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 16724,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 17187,
            "output_tokens": 65,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 17783,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 18035,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18538,
            "output_tokens": 140,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 19990,
            "output_tokens": 326,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20096,
            "input_tokens": 20643,
            "output_tokens": 497,
            "reasoning_output_tokens": 139
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21573,
            "output_tokens": 501,
            "reasoning_output_tokens": 178
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17152,
            "input_tokens": 22603,
            "output_tokens": 295,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 23173,
            "output_tokens": 84,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23952,
            "output_tokens": 1262,
            "reasoning_output_tokens": 177
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25235,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25390,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 25799,
            "output_tokens": 130,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26212,
            "output_tokens": 155,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26636,
            "output_tokens": 153,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26847,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26906,
            "output_tokens": 97,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27432,
            "output_tokens": 441,
            "reasoning_output_tokens": 176
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28120,
            "output_tokens": 273,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28420,
            "output_tokens": 471,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 28914,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 31,
        "model_tool_calls_by_name": {
          "read": 10,
          "ls": 1,
          "bash": 1,
          "exec": 12,
          "devices": 3,
          "write": 2,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 228 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/adjust-bottle.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/workspace/.kinex/memos/adjust-bottle.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/right_tcp_motion.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-11-20261113/workspace/.kinex/tools/right_tcp_motion.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 66,
        "observed_images": 5
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 92,
        "published_frames": 92,
        "decode_ok": true,
        "duration_s": 2.3,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "1fd533727d18286403662d007edda3b4f16fea205c90ee5b5df068170f154eb3",
        "source_sha256": {
          "third_person": "b9192337ef3e965653e3dfc6501d63cdc453ec238765600a743d26417a71bfbe",
          "left_wrist": "dbbfa981439943f6cd5129f942f5dfd8d52d82cd994c5c0876dcefc73b5c474d",
          "right_wrist": "82d16a63b748539ade560174ba21b733a3fab8ef4830a72e5140cade49c1d566"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "3e98c1ab9583",
          "task": "robotwin/adjust-bottle-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 93,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 92,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9120000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 93,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 92,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9120000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 93,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 92,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9120000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-12-20261113",
      "variant": "kinex-native",
      "task_key": "task03/12",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "12",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 11.48,
        "step_limit": 400,
        "steps": 287,
        "success": true,
        "valid": true
      },
      "steps": 287,
      "simulation_time_s": 11.48,
      "wall_time_s": 1037.323749,
      "timing": {
        "simulation_time_s": 11.48,
        "wall_time_s": 1037.323749,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the hammer with black handle and hit the block",
      "native_instruction": "Grab the hammer with black handle and hit the block",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9409123991350667,
        "cache_read_input_tokens": 892032,
        "cache_reported_input_tokens": 948050,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 948050,
        "cached_input_tokens": 892032,
        "cost_usd": null,
        "fresh_input_tokens": 56018,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 948050,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 892032,
        "known_fresh_input_tokens": 56018,
        "known_input_tokens": 948050,
        "known_output_tokens": 7869,
        "known_reasoning_output_tokens": 3564,
        "output_tokens": 7869,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3564,
        "reasoning_reported_output_tokens": 7869,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 955919,
        "uncached_input_tokens": 56018,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11813,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11853,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11986,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14899,
            "output_tokens": 215,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14976,
            "input_tokens": 18043,
            "output_tokens": 196,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 22312,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22574,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 23320,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23572,
            "output_tokens": 116,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24368,
            "output_tokens": 104,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 25511,
            "output_tokens": 87,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25967,
            "output_tokens": 682,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26669,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27443,
            "output_tokens": 125,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 28015,
            "output_tokens": 277,
            "reasoning_output_tokens": 108
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28543,
            "output_tokens": 298,
            "reasoning_output_tokens": 207
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 29285,
            "output_tokens": 208,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29784,
            "output_tokens": 414,
            "reasoning_output_tokens": 280
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30844,
            "output_tokens": 223,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31321,
            "output_tokens": 157,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 32120,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32622,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33363,
            "output_tokens": 1127,
            "reasoning_output_tokens": 1034
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 35171,
            "output_tokens": 107,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35728,
            "output_tokens": 255,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36240,
            "output_tokens": 573,
            "reasoning_output_tokens": 476
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37460,
            "output_tokens": 708,
            "reasoning_output_tokens": 637
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38236,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38787,
            "output_tokens": 240,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39331,
            "output_tokens": 219,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39692,
            "output_tokens": 75,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40164,
            "output_tokens": 826,
            "reasoning_output_tokens": 249
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41014,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 43,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 13,
          "find": 1,
          "exec": 20,
          "devices": 3,
          "write": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 287 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-hammer.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/workspace/.kinex/memos/robotwin-hammer.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-12-20261113/workspace/.kinex/tools/cartesian.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 79,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 116,
        "published_frames": 116,
        "decode_ok": true,
        "duration_s": 2.9,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f86020a8bf23ccc345365be74c2b07dde154d5b695cfb2415ce07daa9efed853",
        "source_sha256": {
          "third_person": "01ce9cc5c31de3db8f33e960a821b0da99a34983582ea34d1189e7b345a08ae5",
          "left_wrist": "bbba70840ce9d97172994a665ee152c75679fa17456984554d7687ad8ba323cb",
          "right_wrist": "ff4be9b7d8f67deb128b16d5f1a8b0ed80d6a5a81b61e0adf49132f69886c913"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "3db81c2a818d",
          "task": "robotwin/beat-block-hammer-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 116,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 116,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11480000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 116,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 116,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11480000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 116,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 116,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11480000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-13-20261113",
      "variant": "kinex-native",
      "task_key": "task03/13",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "13",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 33.76,
        "step_limit": 1200,
        "steps": 844,
        "success": true,
        "valid": true
      },
      "steps": 844,
      "simulation_time_s": 33.76,
      "wall_time_s": 1423.592977,
      "timing": {
        "simulation_time_s": 33.76,
        "wall_time_s": 1423.592977,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set red block on the left, then green block in the center, and finally blue block on the right.",
      "native_instruction": "Set red block on the left, then green block in the center, and finally blue block on the right.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.956335868787168,
        "cache_read_input_tokens": 772224,
        "cache_reported_input_tokens": 807482,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 807482,
        "cached_input_tokens": 772224,
        "cost_usd": null,
        "fresh_input_tokens": 35258,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 807482,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 772224,
        "known_fresh_input_tokens": 35258,
        "known_input_tokens": 807482,
        "known_output_tokens": 6066,
        "known_reasoning_output_tokens": 1201,
        "output_tokens": 6066,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1201,
        "reasoning_reported_output_tokens": 6066,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 31,
          "cache_write_input_tokens": 31,
          "cache_write_reported_input_tokens": 31,
          "cached_input_tokens": 31,
          "fresh_input_tokens": 31,
          "input_tokens": 31,
          "output_tokens": 31,
          "reasoning_output_tokens": 31,
          "reasoning_reported_output_tokens": 31
        },
        "request_attempts": 31,
        "response_count": 31,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 813548,
        "uncached_input_tokens": 35258,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11824,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11864,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11997,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12296,
            "output_tokens": 223,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16967,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 17537,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 18801,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 19031,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 19223,
            "output_tokens": 160,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 21117,
            "output_tokens": 515,
            "reasoning_output_tokens": 138
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 22827,
            "output_tokens": 160,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 24166,
            "output_tokens": 1527,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25714,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26240,
            "output_tokens": 215,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 27054,
            "output_tokens": 119,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27844,
            "output_tokens": 289,
            "reasoning_output_tokens": 138
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28884,
            "output_tokens": 421,
            "reasoning_output_tokens": 193
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29332,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30070,
            "output_tokens": 161,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30291,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30378,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30864,
            "output_tokens": 172,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31729,
            "output_tokens": 245,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32821,
            "output_tokens": 233,
            "reasoning_output_tokens": 100
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33823,
            "output_tokens": 122,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34586,
            "output_tokens": 106,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35344,
            "output_tokens": 183,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35669,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35760,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36384,
            "output_tokens": 637,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37045,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 39,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 14,
          "bash": 1,
          "exec": 15,
          "devices": 2,
          "device": 1,
          "write": 2,
          "edit": 1,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 844 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/blocks-ranking-rgb.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/workspace/.kinex/memos/blocks-ranking-rgb.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/blocks_rgb.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-13-20261113/workspace/.kinex/tools/blocks_rgb.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 73,
        "observed_images": 13
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 339,
        "published_frames": 339,
        "decode_ok": true,
        "duration_s": 8.475,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "89059d4074083a815f61596e15ee66367164778645c4b55b82cf470e6244ac98",
        "source_sha256": {
          "third_person": "4a809369e14c233f01c8aa616904de2e2e2ddc489031eae2a863f82f6f4e68fb",
          "left_wrist": "ebc1b078e4d232d152d2cc432e301510e5e37780ec9deeada68f9046a4021d9c",
          "right_wrist": "07d3ae55dade166b23a85773ea45c3f85b1f43cdc04c02b3187cf18b812e0f3e"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "9c615655a783",
          "task": "robotwin/blocks-ranking-rgb-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 339,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 339,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33760000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 339,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 339,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33760000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 339,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 339,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33760000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-14-20261113",
      "variant": "kinex-native",
      "task_key": "task03/14",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "14",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 4.4,
        "step_limit": 400,
        "steps": 110,
        "success": true,
        "valid": true
      },
      "steps": 110,
      "simulation_time_s": 4.4,
      "wall_time_s": 379.716831,
      "timing": {
        "simulation_time_s": 4.4,
        "wall_time_s": 379.716831,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Click the top button of the alarm-clock with beige screen",
      "native_instruction": "Click the top button of the alarm-clock with beige screen",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9553486170395097,
        "cache_read_input_tokens": 702336,
        "cache_reported_input_tokens": 735162,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 735162,
        "cached_input_tokens": 702336,
        "cost_usd": null,
        "fresh_input_tokens": 32826,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 735162,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 702336,
        "known_fresh_input_tokens": 32826,
        "known_input_tokens": 735162,
        "known_output_tokens": 4617,
        "known_reasoning_output_tokens": 756,
        "output_tokens": 4617,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 756,
        "reasoning_reported_output_tokens": 4617,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 30,
          "cache_write_input_tokens": 30,
          "cache_write_reported_input_tokens": 30,
          "cached_input_tokens": 30,
          "fresh_input_tokens": 30,
          "input_tokens": 30,
          "output_tokens": 30,
          "reasoning_output_tokens": 30,
          "reasoning_reported_output_tokens": 30
        },
        "request_attempts": 30,
        "response_count": 30,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 739779,
        "uncached_input_tokens": 32826,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11814,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11947,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11995,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12243,
            "output_tokens": 223,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 16914,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16768,
            "input_tokens": 18176,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18406,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 19836,
            "output_tokens": 71,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 20117,
            "output_tokens": 264,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 20776,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 21184,
            "output_tokens": 248,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21967,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22842,
            "output_tokens": 163,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 24928,
            "output_tokens": 50,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25041,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25399,
            "output_tokens": 369,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26654,
            "output_tokens": 212,
            "reasoning_output_tokens": 120
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 28509,
            "output_tokens": 723,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29253,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29523,
            "output_tokens": 219,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29930,
            "output_tokens": 216,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30430,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30918,
            "output_tokens": 48,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31136,
            "output_tokens": 203,
            "reasoning_output_tokens": 116
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31750,
            "output_tokens": 118,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32010,
            "output_tokens": 73,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32105,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32717,
            "output_tokens": 174,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33044,
            "output_tokens": 532,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33598,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 40,
        "model_tool_calls_by_name": {
          "read": 16,
          "ls": 1,
          "bash": 1,
          "exec": 12,
          "devices": 5,
          "device": 2,
          "write": 2,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 110 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/workspace/.kinex/memos/robotwin.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-14-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 73,
        "observed_images": 9
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 45,
        "published_frames": 45,
        "decode_ok": true,
        "duration_s": 1.125,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "05f410a8535fd4e3906e437e7287efbf0a19d86a25062892b168d18c18fc4a72",
        "source_sha256": {
          "third_person": "b2c993eea2f69451602ca29de80640045a0cdbd58bb9685155d94a34b5b0b7fa",
          "left_wrist": "11fd18db6e2c65a0c146867d660f3611f479303b2b6312352e1377e250eb08c3",
          "right_wrist": "0182616ba5698dd1c6094ab01c55f60816c84e85874040b3608d32efb06b1f66"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "cec9fa5d66d8",
          "task": "robotwin/click-alarmclock-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 46,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 45,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 4400000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 46,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 45,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 4400000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 46,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 45,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 4400000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-15-20261113",
      "variant": "kinex-native",
      "task_key": "task03/15",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "15",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 8.8,
        "step_limit": 400,
        "steps": 220,
        "success": false,
        "valid": true
      },
      "steps": 220,
      "simulation_time_s": 8.8,
      "wall_time_s": 586.310654,
      "timing": {
        "simulation_time_s": 8.8,
        "wall_time_s": 586.310654,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to press the medium-sized bell's top center.",
      "native_instruction": "Use the left arm to press the medium-sized bell's top center.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9626330793436289,
        "cache_read_input_tokens": 689536,
        "cache_reported_input_tokens": 716302,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 716302,
        "cached_input_tokens": 689536,
        "cost_usd": null,
        "fresh_input_tokens": 26766,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 716302,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 689536,
        "known_fresh_input_tokens": 26766,
        "known_input_tokens": 716302,
        "known_output_tokens": 6356,
        "known_reasoning_output_tokens": 1916,
        "output_tokens": 6356,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1916,
        "reasoning_reported_output_tokens": 6356,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 27,
          "cache_write_input_tokens": 27,
          "cache_write_reported_input_tokens": 27,
          "cached_input_tokens": 27,
          "fresh_input_tokens": 27,
          "input_tokens": 27,
          "output_tokens": 27,
          "reasoning_output_tokens": 27,
          "reasoning_reported_output_tokens": 27
        },
        "request_attempts": 27,
        "response_count": 27,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 722658,
        "uncached_input_tokens": 26766,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11988,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15761,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 18841,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 20103,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20267,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 21695,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 21947,
            "output_tokens": 371,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22713,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23121,
            "output_tokens": 322,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23978,
            "output_tokens": 266,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 25842,
            "output_tokens": 199,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 28377,
            "output_tokens": 461,
            "reasoning_output_tokens": 106
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29385,
            "output_tokens": 508,
            "reasoning_output_tokens": 170
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30360,
            "output_tokens": 944,
            "reasoning_output_tokens": 218
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31327,
            "output_tokens": 160,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31545,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31599,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32101,
            "output_tokens": 186,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32836,
            "output_tokens": 257,
            "reasoning_output_tokens": 138
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33646,
            "output_tokens": 412,
            "reasoning_output_tokens": 294
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34609,
            "output_tokens": 763,
            "reasoning_output_tokens": 654
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 36880,
            "output_tokens": 234,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37669,
            "output_tokens": 72,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37763,
            "output_tokens": 490,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38279,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 35,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 12,
          "find": 1,
          "exec": 13,
          "devices": 2,
          "device": 1,
          "write": 2,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 220 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-bell.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/workspace/.kinex/memos/robotwin-bell.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/bell_press.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-15-20261113/workspace/.kinex/tools/bell_press.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 65,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 89,
        "published_frames": 89,
        "decode_ok": true,
        "duration_s": 2.225,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "2736801c152ad25e605236976b828c9755eb2d6a649a1cd85afaf961b3452f43",
        "source_sha256": {
          "third_person": "e0a537b28d932226bffc4588ae994b1a6b2c98dae56675f21c0654926ad3524a",
          "left_wrist": "bebf5f0597eeafc157d948802afaebf814c078a676101c3250fa067c5f054b9b",
          "right_wrist": "60f8c2eb1ce85133cdc9d60fbc7c529b17c14506fa40ca98cc30246be8667af3"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "a1363c49da93",
          "task": "robotwin/click-bell-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 90,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 89,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8800000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 90,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 89,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8800000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 90,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 89,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8800000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-16-20261113",
      "variant": "kinex-native",
      "task_key": "task03/16",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "16",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 23.92,
        "step_limit": 600,
        "steps": 598,
        "success": false,
        "valid": true
      },
      "steps": 598,
      "simulation_time_s": 23.92,
      "wall_time_s": 2222.375313,
      "timing": {
        "simulation_time_s": 23.92,
        "wall_time_s": 2222.375313,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the compact tabletop trashbin, pour balls into the bin.",
      "native_instruction": "Pick up the compact tabletop trashbin, pour balls into the bin.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9779798052778061,
        "cache_read_input_tokens": 2544768,
        "cache_reported_input_tokens": 2602066,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2602066,
        "cached_input_tokens": 2544768,
        "cost_usd": null,
        "fresh_input_tokens": 57298,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2602066,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2544768,
        "known_fresh_input_tokens": 57298,
        "known_input_tokens": 2602066,
        "known_output_tokens": 15879,
        "known_reasoning_output_tokens": 6785,
        "output_tokens": 15879,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 6785,
        "reasoning_reported_output_tokens": 15879,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 69,
          "cache_write_input_tokens": 69,
          "cache_write_reported_input_tokens": 69,
          "cached_input_tokens": 69,
          "fresh_input_tokens": 69,
          "input_tokens": 69,
          "output_tokens": 69,
          "reasoning_output_tokens": 69,
          "reasoning_reported_output_tokens": 69
        },
        "request_attempts": 69,
        "response_count": 69,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2617945,
        "uncached_input_tokens": 57298,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11948,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11996,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12422,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16948,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 18223,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 18453,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 18644,
            "output_tokens": 94,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 19269,
            "output_tokens": 122,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 20834,
            "output_tokens": 169,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 22205,
            "output_tokens": 108,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 22899,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23307,
            "output_tokens": 260,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24037,
            "output_tokens": 330,
            "reasoning_output_tokens": 288
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25546,
            "output_tokens": 805,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26372,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26536,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27023,
            "output_tokens": 126,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27508,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27915,
            "output_tokens": 115,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28392,
            "output_tokens": 94,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 29181,
            "output_tokens": 625,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30079,
            "output_tokens": 207,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30970,
            "output_tokens": 442,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31705,
            "output_tokens": 401,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32605,
            "output_tokens": 936,
            "reasoning_output_tokens": 510
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 34392,
            "output_tokens": 524,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35608,
            "output_tokens": 266,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 35934,
            "output_tokens": 48,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36044,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36528,
            "output_tokens": 246,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37135,
            "output_tokens": 72,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37978,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38019,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38211,
            "output_tokens": 321,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38923,
            "output_tokens": 400,
            "reasoning_output_tokens": 256
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 40116,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40208,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40788,
            "output_tokens": 215,
            "reasoning_output_tokens": 174
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41379,
            "output_tokens": 112,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41518,
            "output_tokens": 458,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42365,
            "output_tokens": 459,
            "reasoning_output_tokens": 323
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43508,
            "output_tokens": 183,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 44057,
            "output_tokens": 195,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44613,
            "output_tokens": 244,
            "reasoning_output_tokens": 219
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45026,
            "output_tokens": 1755,
            "reasoning_output_tokens": 1552
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 47006,
            "output_tokens": 145,
            "reasoning_output_tokens": 6
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 47514,
            "output_tokens": 66,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47488,
            "input_tokens": 47602,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47488,
            "input_tokens": 47795,
            "output_tokens": 677,
            "reasoning_output_tokens": 500
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 49156,
            "output_tokens": 175,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49152,
            "input_tokens": 49694,
            "output_tokens": 89,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49664,
            "input_tokens": 50145,
            "output_tokens": 177,
            "reasoning_output_tokens": 120
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50176,
            "input_tokens": 50683,
            "output_tokens": 284,
            "reasoning_output_tokens": 185
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50816,
            "input_tokens": 51327,
            "output_tokens": 114,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51200,
            "input_tokens": 51805,
            "output_tokens": 320,
            "reasoning_output_tokens": 96
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52096,
            "input_tokens": 52405,
            "output_tokens": 336,
            "reasoning_output_tokens": 199
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52608,
            "input_tokens": 52912,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52864,
            "input_tokens": 53770,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53760,
            "input_tokens": 53827,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53760,
            "input_tokens": 54245,
            "output_tokens": 203,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54400,
            "input_tokens": 55214,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55168,
            "input_tokens": 55270,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55296,
            "input_tokens": 55849,
            "output_tokens": 388,
            "reasoning_output_tokens": 140
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56064,
            "input_tokens": 56264,
            "output_tokens": 73,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56192,
            "input_tokens": 56699,
            "output_tokens": 131,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56704,
            "input_tokens": 57192,
            "output_tokens": 981,
            "reasoning_output_tokens": 354
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57984,
            "input_tokens": 58198,
            "output_tokens": 52,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58112,
            "input_tokens": 58312,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 79,
        "model_tool_calls_by_name": {
          "read": 17,
          "ls": 1,
          "find": 4,
          "exec": 37,
          "devices": 5,
          "device": 5,
          "write": 2,
          "device_stop": 6,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 598 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-bin-pouring.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/workspace/.kinex/memos/robotwin-bin-pouring.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-16-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 151,
        "observed_images": 25
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 240,
        "published_frames": 240,
        "decode_ok": true,
        "duration_s": 6.0,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "cfeb93605057f83611334fc8d5360ac47cfc9cdb8ddc40be1e700dba37893a10",
        "source_sha256": {
          "third_person": "69bb93756632b64a146af46a77d6adc981fbf71345810c61f910a19f20cf6153",
          "left_wrist": "b4e4eb9663df2f3e61f041b22dbc2634374a1fb319ec9df9e9b2ad21e3e0c4b0",
          "right_wrist": "f80063a0bc61c043e62c2fb9e7ddc039d1aab8afb9fb0b198c4fa373c20a5032"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "39100c7994b3",
          "task": "robotwin/dump-bin-bigbin-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 241,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 240,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23920000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 241,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 240,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23920000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 241,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 240,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23920000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-17-20261113",
      "variant": "kinex-native",
      "task_key": "task03/17",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "17",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 7.48,
        "step_limit": 400,
        "steps": 187,
        "success": true,
        "valid": true
      },
      "steps": 187,
      "simulation_time_s": 7.48,
      "wall_time_s": 552.344094,
      "timing": {
        "simulation_time_s": 7.48,
        "wall_time_s": 552.344094,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use both arms to grab the light brown wooden roller.",
      "native_instruction": "Use both arms to grab the light brown wooden roller.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9238031570187047,
        "cache_read_input_tokens": 677120,
        "cache_reported_input_tokens": 732970,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 732970,
        "cached_input_tokens": 677120,
        "cost_usd": null,
        "fresh_input_tokens": 55850,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 732970,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 677120,
        "known_fresh_input_tokens": 55850,
        "known_input_tokens": 732970,
        "known_output_tokens": 5164,
        "known_reasoning_output_tokens": 825,
        "output_tokens": 5164,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 825,
        "reasoning_reported_output_tokens": 5164,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 28,
          "cache_write_input_tokens": 28,
          "cache_write_reported_input_tokens": 28,
          "cached_input_tokens": 28,
          "fresh_input_tokens": 28,
          "input_tokens": 28,
          "output_tokens": 28,
          "reasoning_output_tokens": 28,
          "reasoning_reported_output_tokens": 28
        },
        "request_attempts": 28,
        "response_count": 28,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 738134,
        "uncached_input_tokens": 55850,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11988,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 14391,
            "output_tokens": 213,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 17502,
            "output_tokens": 175,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 21610,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 21840,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22031,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 22283,
            "output_tokens": 136,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23225,
            "output_tokens": 54,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23648,
            "output_tokens": 457,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24706,
            "output_tokens": 160,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 27202,
            "output_tokens": 475,
            "reasoning_output_tokens": 147
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28149,
            "output_tokens": 1008,
            "reasoning_output_tokens": 149
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29180,
            "output_tokens": 149,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29828,
            "output_tokens": 54,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29944,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30055,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30705,
            "output_tokens": 270,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31615,
            "output_tokens": 65,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31742,
            "output_tokens": 268,
            "reasoning_output_tokens": 124
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32658,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33351,
            "output_tokens": 177,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33665,
            "output_tokens": 50,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33777,
            "output_tokens": 170,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34301,
            "output_tokens": 376,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34706,
            "output_tokens": 469,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35198,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 38,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 14,
          "find": 1,
          "exec": 11,
          "devices": 1,
          "device": 1,
          "write": 2,
          "device_stop": 3,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 187 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/roller.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/workspace/.kinex/memos/roller.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-17-20261113/workspace/.kinex/tools/dual_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 69,
        "observed_images": 6
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 76,
        "published_frames": 76,
        "decode_ok": true,
        "duration_s": 1.9,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "a2bdab6ee587245f14e932422762202ad3b7b8a50e3afbbe4de827ced1b7d544",
        "source_sha256": {
          "third_person": "c9aa75f40cae46828028cc3593c5723bb2d26ce75e45a05a037b5368d51f881f",
          "left_wrist": "e1b341e7fde324d3698dc283a2f36cdf9828399b592e7e7d6870e7c6131d9e9e",
          "right_wrist": "f2faddc3fe7b2ab2715fb732c6c5e53e8355bf3686fd8291cb88bc8a7e7ef90d"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "1c1738be1664",
          "task": "robotwin/grab-roller-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 76,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 76,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7480000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 76,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 76,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7480000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 76,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 76,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7480000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-18-20261113",
      "variant": "kinex-native",
      "task_key": "task03/18",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "18",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 18.68,
        "step_limit": 600,
        "steps": 467,
        "success": false,
        "valid": true
      },
      "steps": 467,
      "simulation_time_s": 18.68,
      "wall_time_s": 1110.061506,
      "timing": {
        "simulation_time_s": 18.68,
        "wall_time_s": 1110.061506,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the microphone used for audio input and pass it to the other arm.",
      "native_instruction": "Grab the microphone used for audio input and pass it to the other arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9579571021564144,
        "cache_read_input_tokens": 1247104,
        "cache_reported_input_tokens": 1301837,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1301837,
        "cached_input_tokens": 1247104,
        "cost_usd": null,
        "fresh_input_tokens": 54733,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1301837,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1247104,
        "known_fresh_input_tokens": 54733,
        "known_input_tokens": 1301837,
        "known_output_tokens": 8305,
        "known_reasoning_output_tokens": 2401,
        "output_tokens": 8305,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2401,
        "reasoning_reported_output_tokens": 8305,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 42,
          "cache_write_input_tokens": 42,
          "cache_write_reported_input_tokens": 42,
          "cached_input_tokens": 42,
          "fresh_input_tokens": 42,
          "input_tokens": 42,
          "output_tokens": 42,
          "reasoning_output_tokens": 42,
          "reasoning_reported_output_tokens": 42
        },
        "request_attempts": 42,
        "response_count": 42,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1310142,
        "uncached_input_tokens": 54733,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11817,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11857,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11988,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15804,
            "output_tokens": 246,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 19742,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20570,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20800,
            "output_tokens": 132,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 22420,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22672,
            "output_tokens": 382,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23449,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23857,
            "output_tokens": 303,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24695,
            "output_tokens": 296,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25723,
            "output_tokens": 115,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 27481,
            "output_tokens": 1234,
            "reasoning_output_tokens": 389
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28736,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28850,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29441,
            "output_tokens": 53,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29556,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29649,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30176,
            "output_tokens": 144,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30827,
            "output_tokens": 125,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31444,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32053,
            "output_tokens": 302,
            "reasoning_output_tokens": 221
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 33895,
            "output_tokens": 379,
            "reasoning_output_tokens": 277
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34352,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34954,
            "output_tokens": 253,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35564,
            "output_tokens": 271,
            "reasoning_output_tokens": 169
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36531,
            "output_tokens": 279,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37006,
            "output_tokens": 248,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37591,
            "output_tokens": 274,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38379,
            "output_tokens": 209,
            "reasoning_output_tokens": 106
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 39080,
            "output_tokens": 50,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39192,
            "output_tokens": 324,
            "reasoning_output_tokens": 222
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 40012,
            "output_tokens": 125,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40633,
            "output_tokens": 240,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41170,
            "output_tokens": 207,
            "reasoning_output_tokens": 113
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41874,
            "output_tokens": 77,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42432,
            "output_tokens": 74,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42989,
            "output_tokens": 298,
            "reasoning_output_tokens": 158
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43792,
            "output_tokens": 204,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44023,
            "output_tokens": 713,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 44761,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 53,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 15,
          "find": 1,
          "devices": 1,
          "device": 1,
          "exec": 25,
          "write": 2,
          "device_stop": 2,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 467 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/handover-microphone.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/workspace/.kinex/memos/handover-microphone.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-18-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 98,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 188,
        "published_frames": 188,
        "decode_ok": true,
        "duration_s": 4.7,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "77ee1e16411b49c62d3a4ed21089da9cc3ec336638bab08c7f1a2467d7a6b4c7",
        "source_sha256": {
          "third_person": "84702514422533ee50fb03d5ac4b1d4e7148c76a2c2f1ae433f44655c4ee0b49",
          "left_wrist": "3f02748e67eb70535db6ed00b54f9f8a44b5bb0c62e0a8097d571276d27b195f",
          "right_wrist": "8299f476230cd19b11374b42e0cbd6a85b5fa96ccf03465153ffed639813d96b"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "47c3948cc56a",
          "task": "robotwin/handover-mic-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 188,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 188,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18680000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 188,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 188,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18680000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 188,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 188,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18680000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-19-20261113",
      "variant": "kinex-native",
      "task_key": "task03/19",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "19",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.6,
        "step_limit": 400,
        "steps": 240,
        "success": true,
        "valid": true
      },
      "steps": 240,
      "simulation_time_s": 9.6,
      "wall_time_s": 620.349269,
      "timing": {
        "simulation_time_s": 9.6,
        "wall_time_s": 620.349269,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the medium sauce can and place it near the smooth gray pot with matte finish",
      "native_instruction": "Grab the medium sauce can and place it near the smooth gray pot with matte finish",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.967143090556848,
        "cache_read_input_tokens": 994432,
        "cache_reported_input_tokens": 1028216,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1028216,
        "cached_input_tokens": 994432,
        "cost_usd": null,
        "fresh_input_tokens": 33784,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1028216,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 994432,
        "known_fresh_input_tokens": 33784,
        "known_input_tokens": 1028216,
        "known_output_tokens": 6430,
        "known_reasoning_output_tokens": 899,
        "output_tokens": 6430,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 899,
        "reasoning_reported_output_tokens": 6430,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 34,
          "cache_write_input_tokens": 34,
          "cache_write_reported_input_tokens": 34,
          "cached_input_tokens": 34,
          "fresh_input_tokens": 34,
          "input_tokens": 34,
          "output_tokens": 34,
          "reasoning_output_tokens": 34,
          "reasoning_reported_output_tokens": 34
        },
        "request_attempts": 34,
        "response_count": 34,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1034646,
        "uncached_input_tokens": 33784,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11822,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11862,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11995,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14710,
            "output_tokens": 213,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14720,
            "input_tokens": 17821,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 20351,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 20921,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21151,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21341,
            "output_tokens": 175,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 23621,
            "output_tokens": 300,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24456,
            "output_tokens": 272,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 26008,
            "output_tokens": 199,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 29312,
            "output_tokens": 230,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30514,
            "output_tokens": 383,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31097,
            "output_tokens": 505,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32211,
            "output_tokens": 376,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32700,
            "output_tokens": 69,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33581,
            "output_tokens": 348,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34204,
            "output_tokens": 341,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34848,
            "output_tokens": 983,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35854,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36433,
            "output_tokens": 108,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 37051,
            "output_tokens": 39,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37108,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37302,
            "output_tokens": 101,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37918,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38395,
            "output_tokens": 258,
            "reasoning_output_tokens": 185
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39164,
            "output_tokens": 85,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39758,
            "output_tokens": 113,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40385,
            "output_tokens": 83,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40610,
            "output_tokens": 70,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40702,
            "output_tokens": 137,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41188,
            "output_tokens": 607,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41822,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 45,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 16,
          "find": 1,
          "exec": 18,
          "devices": 1,
          "device": 3,
          "write": 2,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 240 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-19-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 82,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 97,
        "published_frames": 97,
        "decode_ok": true,
        "duration_s": 2.425,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f1c1db6d485040c5529e4e92d85755a8c780bf1ace2b3a6daa84b5769a773e5f",
        "source_sha256": {
          "third_person": "a2b644d31e838c513c4e255fb853ebd13914f44ade4d69f2b262ef7d83599501",
          "left_wrist": "c7a06b831de9f67fb0dcedcdcebbbd6b341270ff2ddf12cf0506dede9b65aec8",
          "right_wrist": "9bf5039ebccb7040bb55e27edc52e6dfd3f6b0530d5790befb6da509c597b7e1"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "456068d0b5f0",
          "task": "robotwin/move-can-pot-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 97,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 97,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 97,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-20-20261113",
      "variant": "kinex-native",
      "task_key": "task03/20",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "20",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 10.16,
        "step_limit": 400,
        "steps": 254,
        "success": true,
        "valid": true
      },
      "steps": 254,
      "simulation_time_s": 10.16,
      "wall_time_s": 621.207353,
      "timing": {
        "simulation_time_s": 10.16,
        "wall_time_s": 621.207353,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the medium-sized bottle and put it onto the pad",
      "native_instruction": "Pick the medium-sized bottle and put it onto the pad",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.954044807794401,
        "cache_read_input_tokens": 724224,
        "cache_reported_input_tokens": 759109,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 759109,
        "cached_input_tokens": 724224,
        "cost_usd": null,
        "fresh_input_tokens": 34885,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 759109,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 724224,
        "known_fresh_input_tokens": 34885,
        "known_input_tokens": 759109,
        "known_output_tokens": 4635,
        "known_reasoning_output_tokens": 864,
        "output_tokens": 4635,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 864,
        "reasoning_reported_output_tokens": 4635,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 30,
          "cache_write_input_tokens": 30,
          "cache_write_reported_input_tokens": 30,
          "cached_input_tokens": 30,
          "fresh_input_tokens": 30,
          "input_tokens": 30,
          "output_tokens": 30,
          "reasoning_output_tokens": 30,
          "reasoning_reported_output_tokens": 30
        },
        "request_attempts": 30,
        "response_count": 30,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 763744,
        "uncached_input_tokens": 34885,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11991,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12039,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14724,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14720,
            "input_tokens": 18609,
            "output_tokens": 73,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 19409,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 19979,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 21033,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21263,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21454,
            "output_tokens": 171,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 22629,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22986,
            "output_tokens": 316,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 25094,
            "output_tokens": 195,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26932,
            "output_tokens": 577,
            "reasoning_output_tokens": 239
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 28095,
            "output_tokens": 759,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28875,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29077,
            "output_tokens": 72,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29844,
            "output_tokens": 211,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30388,
            "output_tokens": 183,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30911,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31413,
            "output_tokens": 153,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31990,
            "output_tokens": 79,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32530,
            "output_tokens": 137,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33083,
            "output_tokens": 242,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33738,
            "output_tokens": 61,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33941,
            "output_tokens": 60,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34063,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34370,
            "output_tokens": 579,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 34973,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 37,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 13,
          "find": 1,
          "exec": 13,
          "devices": 4,
          "device": 1,
          "write": 2,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 254 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin_bottle.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/workspace/.kinex/memos/robotwin_bottle.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-20-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 70,
        "observed_images": 9
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 103,
        "published_frames": 103,
        "decode_ok": true,
        "duration_s": 2.575,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f1df43df6d63408ac49b96e3d8e9119c99c511127b1e81f19c1f572476e0a70e",
        "source_sha256": {
          "third_person": "400caa9934ee4261c923d9a1edf3ec25192da8f24eb41ae16faf0c435a63f24d",
          "left_wrist": "44ae9849708f879e90d885086a69e84e5ec1befee966ef9a32fdf1d1af738b1d",
          "right_wrist": "7a4ddc736aac21c6dfd6088f04143cd43885237abbeb4f47870921ae5863552c"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "f04907b36cf9",
          "task": "robotwin/move-pillbottle-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 103,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 103,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10160000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 103,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 103,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10160000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 103,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 103,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10160000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-21-20261113",
      "variant": "kinex-native",
      "task_key": "task03/21",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "21",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 8.08,
        "step_limit": 400,
        "steps": 202,
        "success": false,
        "valid": true
      },
      "steps": 202,
      "simulation_time_s": 8.08,
      "wall_time_s": 535.383589,
      "timing": {
        "simulation_time_s": 8.08,
        "wall_time_s": 535.383589,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the small packet of playingcards and move it outward.",
      "native_instruction": "Pick up the small packet of playingcards and move it outward.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9562984059890896,
        "cache_read_input_tokens": 813568,
        "cache_reported_input_tokens": 850747,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 850747,
        "cached_input_tokens": 813568,
        "cost_usd": null,
        "fresh_input_tokens": 37179,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 850747,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 813568,
        "known_fresh_input_tokens": 37179,
        "known_input_tokens": 850747,
        "known_output_tokens": 5261,
        "known_reasoning_output_tokens": 921,
        "output_tokens": 5261,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 921,
        "reasoning_reported_output_tokens": 5261,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 31,
          "cache_write_input_tokens": 31,
          "cache_write_reported_input_tokens": 31,
          "cached_input_tokens": 31,
          "fresh_input_tokens": 31,
          "input_tokens": 31,
          "output_tokens": 31,
          "reasoning_output_tokens": 31,
          "reasoning_reported_output_tokens": 31
        },
        "request_attempts": 31,
        "response_count": 31,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 856008,
        "uncached_input_tokens": 37179,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11817,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11857,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11990,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 16089,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 16208,
            "output_tokens": 192,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16256,
            "input_tokens": 20059,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 21322,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21892,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22088,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22279,
            "output_tokens": 192,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 24530,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24572,
            "output_tokens": 320,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 26127,
            "output_tokens": 207,
            "reasoning_output_tokens": 167
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26800,
            "output_tokens": 245,
            "reasoning_output_tokens": 203
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 28224,
            "output_tokens": 456,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28987,
            "output_tokens": 355,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29877,
            "output_tokens": 287,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30633,
            "output_tokens": 859,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31513,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32086,
            "output_tokens": 134,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 33176,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33367,
            "output_tokens": 188,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33824,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34707,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34898,
            "output_tokens": 276,
            "reasoning_output_tokens": 181
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 36210,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36404,
            "output_tokens": 365,
            "reasoning_output_tokens": 131
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36796,
            "output_tokens": 69,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36897,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37516,
            "output_tokens": 462,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38002,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 38,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 15,
          "find": 1,
          "bash": 1,
          "exec": 10,
          "devices": 1,
          "device": 5,
          "write": 2,
          "edit": 1,
          "grep": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 202 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-playingcards.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/workspace/.kinex/memos/robotwin-playingcards.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-21-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 72,
        "observed_images": 5
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 82,
        "published_frames": 82,
        "decode_ok": true,
        "duration_s": 2.05,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "da91fe084e97f1ef07fe58af11183a7e3421ded46bf0c5c50eb018ee4c790117",
        "source_sha256": {
          "third_person": "6592b3359cdaee9526b0cc90e263d9ded7a7f4a3a7c3b806b7bbd08d415f8cf8",
          "left_wrist": "460de9bb21b4e5f504cd5da0f36d645276cbf80a3470478ed6d2930b817f10c6",
          "right_wrist": "ab706b114510b2c25ee5a46d3e848fa19ef43a2f15ba15fea4314f3ab635798e"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "fdd469189f96",
          "task": "robotwin/move-playingcard-away-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 82,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 82,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8080000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 82,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 82,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8080000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 82,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 82,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8080000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-22-20261113",
      "variant": "kinex-native",
      "task_key": "task03/22",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "22",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 10.52,
        "step_limit": 400,
        "steps": 263,
        "success": false,
        "valid": true
      },
      "steps": 263,
      "simulation_time_s": 10.52,
      "wall_time_s": 968.747277,
      "timing": {
        "simulation_time_s": 10.52,
        "wall_time_s": 968.747277,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the red plastic stapler and set it on Red mat",
      "native_instruction": "Grab the red plastic stapler and set it on Red mat",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9786952978943458,
        "cache_read_input_tokens": 1293568,
        "cache_reported_input_tokens": 1321727,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1321727,
        "cached_input_tokens": 1293568,
        "cost_usd": null,
        "fresh_input_tokens": 28159,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1321727,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1293568,
        "known_fresh_input_tokens": 28159,
        "known_input_tokens": 1321727,
        "known_output_tokens": 6506,
        "known_reasoning_output_tokens": 1200,
        "output_tokens": 6506,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1200,
        "reasoning_reported_output_tokens": 6506,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 48,
          "cache_write_input_tokens": 48,
          "cache_write_reported_input_tokens": 48,
          "cached_input_tokens": 48,
          "fresh_input_tokens": 48,
          "input_tokens": 48,
          "output_tokens": 48,
          "reasoning_output_tokens": 48,
          "reasoning_reported_output_tokens": 48
        },
        "request_attempts": 48,
        "response_count": 48,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1328233,
        "uncached_input_tokens": 28159,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11816,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11856,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11989,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12288,
            "output_tokens": 192,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 16139,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 17402,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 17566,
            "output_tokens": 139,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 19557,
            "output_tokens": 95,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 19862,
            "output_tokens": 188,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 21464,
            "output_tokens": 266,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 22180,
            "output_tokens": 215,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23622,
            "output_tokens": 62,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24475,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24577,
            "output_tokens": 324,
            "reasoning_output_tokens": 165
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25436,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25687,
            "output_tokens": 296,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26730,
            "output_tokens": 880,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27631,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27757,
            "output_tokens": 58,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 27877,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28364,
            "output_tokens": 175,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28664,
            "output_tokens": 66,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29099,
            "output_tokens": 157,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29347,
            "output_tokens": 196,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29570,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29708,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29799,
            "output_tokens": 356,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30182,
            "output_tokens": 180,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30569,
            "output_tokens": 92,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30763,
            "output_tokens": 41,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30866,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31293,
            "output_tokens": 297,
            "reasoning_output_tokens": 213
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31758,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32138,
            "output_tokens": 79,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32383,
            "output_tokens": 198,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32743,
            "output_tokens": 130,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33042,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33505,
            "output_tokens": 186,
            "reasoning_output_tokens": 100
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33857,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34233,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34483,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34864,
            "output_tokens": 65,
            "reasoning_output_tokens": 6
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35174,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35271,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35526,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35904,
            "output_tokens": 202,
            "reasoning_output_tokens": 92
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36133,
            "output_tokens": 420,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36578,
            "output_tokens": 18,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 55,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 13,
          "bash": 1,
          "exec": 20,
          "devices": 9,
          "device": 3,
          "write": 2,
          "device_stop": 3,
          "edit": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 263 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-22-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 106,
        "observed_images": 9
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 106,
        "published_frames": 106,
        "decode_ok": true,
        "duration_s": 2.65,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f8772510ea0aefd204c32230e68482701abdffd143bbef279a28a60b3725f98c",
        "source_sha256": {
          "third_person": "f11cb2066d05a84f6897083641b855f31e9e932d4dc7764f0c330b4881347249",
          "left_wrist": "0f6c05de36b3cbbb4efc83866e8fb3ecf65da0cf6302109ee16c201190238ef3",
          "right_wrist": "51f13c08ac00fa3b672abee85e5e3bea0736aa77b39a7d696a822f585ca41300"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "e302c19fc48e",
          "task": "robotwin/move-stapler-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 107,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 106,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10520000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 107,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 106,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10520000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 107,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 106,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10520000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-23-20261113",
      "variant": "kinex-native",
      "task_key": "task03/23",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "23",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 8.68,
        "step_limit": 700,
        "steps": 217,
        "success": true,
        "valid": true
      },
      "steps": 217,
      "simulation_time_s": 8.68,
      "wall_time_s": 580.357481,
      "timing": {
        "simulation_time_s": 8.68,
        "wall_time_s": 580.357481,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift the black laptop using the right arm.",
      "native_instruction": "Lift the black laptop using the right arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9700409615715054,
        "cache_read_input_tokens": 700032,
        "cache_reported_input_tokens": 721652,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 721652,
        "cached_input_tokens": 700032,
        "cost_usd": null,
        "fresh_input_tokens": 21620,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 721652,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 700032,
        "known_fresh_input_tokens": 21620,
        "known_input_tokens": 721652,
        "known_output_tokens": 6042,
        "known_reasoning_output_tokens": 1255,
        "output_tokens": 6042,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1255,
        "reasoning_reported_output_tokens": 6042,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 31,
          "cache_write_input_tokens": 31,
          "cache_write_reported_input_tokens": 31,
          "cached_input_tokens": 31,
          "fresh_input_tokens": 31,
          "input_tokens": 31,
          "output_tokens": 31,
          "reasoning_output_tokens": 31,
          "reasoning_reported_output_tokens": 31
        },
        "request_attempts": 31,
        "response_count": 31,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 727694,
        "uncached_input_tokens": 21620,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11810,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11941,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11989,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12237,
            "output_tokens": 162,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 15280,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15232,
            "input_tokens": 16542,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 16706,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 17165,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17024,
            "input_tokens": 17735,
            "output_tokens": 172,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 19350,
            "output_tokens": 317,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 20062,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20470,
            "output_tokens": 317,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 22549,
            "output_tokens": 454,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23315,
            "output_tokens": 615,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24373,
            "output_tokens": 297,
            "reasoning_output_tokens": 255
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25849,
            "output_tokens": 814,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26683,
            "output_tokens": 105,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26853,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27039,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27105,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27521,
            "output_tokens": 138,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27764,
            "output_tokens": 172,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 27962,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28453,
            "output_tokens": 222,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29075,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29494,
            "output_tokens": 516,
            "reasoning_output_tokens": 335
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30289,
            "output_tokens": 140,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30858,
            "output_tokens": 201,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31201,
            "output_tokens": 77,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31673,
            "output_tokens": 614,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32309,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 36,
        "model_tool_calls_by_name": {
          "read": 12,
          "ls": 1,
          "bash": 1,
          "exec": 13,
          "devices": 4,
          "write": 2,
          "device_stop": 1,
          "relative_pose": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 217 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/open-laptop.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/workspace/.kinex/memos/open-laptop.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/right_tcp.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-23-20261113/workspace/.kinex/tools/right_tcp.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 70,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 88,
        "published_frames": 88,
        "decode_ok": true,
        "duration_s": 2.2,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "ec371de414a21918eb8956584c18bf17ccf1e0cfa52d7738a3425012ffe92d00",
        "source_sha256": {
          "third_person": "d37961ded1215ad831c99d118ab233663c373fa70dc82d058c950ea549076b0c",
          "left_wrist": "0d7810b5ff596d58688f9cbccc6c00922554a8481343c76dd01c60d815794e20",
          "right_wrist": "b08ec0ac50519def8dc31e8a0ba1368b14b8198c692c637190836273a098fb53"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "1823778dc7ad",
          "task": "robotwin/open-laptop-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 88,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 88,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8680000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 88,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 88,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8680000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 88,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 88,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8680000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-24-20261113",
      "variant": "kinex-native",
      "task_key": "task03/24",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "24",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 16.64,
        "step_limit": 1500,
        "steps": 416,
        "success": true,
        "valid": true
      },
      "steps": 416,
      "simulation_time_s": 16.64,
      "wall_time_s": 749.82202,
      "timing": {
        "simulation_time_s": 16.64,
        "wall_time_s": 749.82202,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to pull the handle of the kitchen microwave with slanted shape.",
      "native_instruction": "Use the left arm to pull the handle of the kitchen microwave with slanted shape.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9757696788737756,
        "cache_read_input_tokens": 1107200,
        "cache_reported_input_tokens": 1134694,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1134694,
        "cached_input_tokens": 1107200,
        "cost_usd": null,
        "fresh_input_tokens": 27494,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1134694,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1107200,
        "known_fresh_input_tokens": 27494,
        "known_input_tokens": 1134694,
        "known_output_tokens": 5718,
        "known_reasoning_output_tokens": 905,
        "output_tokens": 5718,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 905,
        "reasoning_reported_output_tokens": 5718,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 42,
          "cache_write_input_tokens": 42,
          "cache_write_reported_input_tokens": 42,
          "cached_input_tokens": 42,
          "fresh_input_tokens": 42,
          "input_tokens": 42,
          "output_tokens": 42,
          "reasoning_output_tokens": 42,
          "reasoning_reported_output_tokens": 42
        },
        "request_attempts": 42,
        "response_count": 42,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1140412,
        "uncached_input_tokens": 27494,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11989,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14523,
            "output_tokens": 199,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14592,
            "input_tokens": 17620,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 18882,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19046,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 20476,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20728,
            "output_tokens": 289,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21412,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21820,
            "output_tokens": 261,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22616,
            "output_tokens": 246,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23506,
            "output_tokens": 168,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 26086,
            "output_tokens": 810,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26917,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27469,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27600,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27695,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 27906,
            "output_tokens": 52,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28653,
            "output_tokens": 91,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29113,
            "output_tokens": 53,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29188,
            "output_tokens": 237,
            "reasoning_output_tokens": 146
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29452,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29680,
            "output_tokens": 140,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29955,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30411,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31099,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31259,
            "output_tokens": 50,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31331,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31748,
            "output_tokens": 48,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31814,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32012,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32152,
            "output_tokens": 179,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32696,
            "output_tokens": 55,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32922,
            "output_tokens": 70,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33014,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33438,
            "output_tokens": 188,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34002,
            "output_tokens": 245,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34389,
            "output_tokens": 70,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34855,
            "output_tokens": 700,
            "reasoning_output_tokens": 116
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35578,
            "output_tokens": 361,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35966,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 51,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 13,
          "find": 1,
          "exec": 17,
          "devices": 4,
          "device": 3,
          "write": 2,
          "relative_pose": 2,
          "edit": 3,
          "device_stop": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 416 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/open-microwave.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/workspace/.kinex/memos/open-microwave.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-24-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 96,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1500,
        "max_control_steps": 1500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 168,
        "published_frames": 168,
        "decode_ok": true,
        "duration_s": 4.2,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "035e60422833acbdde71638f733132208c5b9f5e422d409fc2605eac60701714",
        "source_sha256": {
          "third_person": "5ec1843e2fc4fa079823f293c89a1598e8b6c6432e40b8889164e6fc59e5d743",
          "left_wrist": "ae1af5d0f29de74e7a3c2682ac645bf623fb6db660271becb7ee8157537aceed",
          "right_wrist": "0615d123cb9043c19072290424fb57a9081408a34381002247491f5a76cfb18c"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "887f34498252",
          "task": "robotwin/open-microwave-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 168,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 168,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 168,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 168,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 168,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 168,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-25-20261113",
      "variant": "kinex-native",
      "task_key": "task03/25",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "25",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 2
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.56,
        "step_limit": 400,
        "steps": 389,
        "success": false,
        "valid": true
      },
      "steps": 389,
      "simulation_time_s": 15.56,
      "wall_time_s": 1341.169659,
      "timing": {
        "simulation_time_s": 15.56,
        "wall_time_s": 1341.169659,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift the bottle with attached loop on cap and the plastic bottle with blue label simultaneously with separate arms.",
      "native_instruction": "Lift the bottle with attached loop on cap and the plastic bottle with blue label simultaneously with separate arms.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9578516380559102,
        "cache_read_input_tokens": 1585664,
        "cache_reported_input_tokens": 1655438,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1655438,
        "cached_input_tokens": 1585664,
        "cost_usd": null,
        "fresh_input_tokens": 69774,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1655438,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1585664,
        "known_fresh_input_tokens": 69774,
        "known_input_tokens": 1655438,
        "known_output_tokens": 10682,
        "known_reasoning_output_tokens": 3405,
        "output_tokens": 10682,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3405,
        "reasoning_reported_output_tokens": 10682,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 54,
          "cache_write_input_tokens": 54,
          "cache_write_reported_input_tokens": 54,
          "cached_input_tokens": 54,
          "fresh_input_tokens": 54,
          "input_tokens": 54,
          "output_tokens": 54,
          "reasoning_output_tokens": 54,
          "reasoning_reported_output_tokens": 54
        },
        "request_attempts": 54,
        "response_count": 54,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1666120,
        "uncached_input_tokens": 69774,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11826,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11959,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 12011,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12259,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12722,
            "output_tokens": 195,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12800,
            "input_tokens": 16936,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 18198,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 18428,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 18619,
            "output_tokens": 88,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 19238,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 19490,
            "output_tokens": 139,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 20942,
            "output_tokens": 345,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21822,
            "output_tokens": 328,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22879,
            "output_tokens": 147,
            "reasoning_output_tokens": 108
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23721,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24225,
            "output_tokens": 502,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24892,
            "output_tokens": 128,
            "reasoning_output_tokens": 86
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 26199,
            "output_tokens": 929,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27150,
            "output_tokens": 105,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27639,
            "output_tokens": 195,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28331,
            "output_tokens": 47,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28440,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28524,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 29121,
            "output_tokens": 186,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 30013,
            "output_tokens": 120,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30498,
            "output_tokens": 70,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 30630,
            "output_tokens": 160,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 32330,
            "output_tokens": 96,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32489,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 32850,
            "output_tokens": 466,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33589,
            "output_tokens": 481,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34351,
            "output_tokens": 291,
            "reasoning_output_tokens": 196
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 35175,
            "output_tokens": 38,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35384,
            "output_tokens": 223,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35815,
            "output_tokens": 131,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36167,
            "output_tokens": 145,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36372,
            "output_tokens": 47,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36481,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 37061,
            "output_tokens": 110,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37537,
            "output_tokens": 583,
            "reasoning_output_tokens": 329
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38531,
            "output_tokens": 55,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38756,
            "output_tokens": 471,
            "reasoning_output_tokens": 395
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39915,
            "output_tokens": 134,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40737,
            "output_tokens": 371,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41263,
            "output_tokens": 125,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41924,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42356,
            "output_tokens": 616,
            "reasoning_output_tokens": 480
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43677,
            "output_tokens": 205,
            "reasoning_output_tokens": 164
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 43904,
            "output_tokens": 365,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44867,
            "output_tokens": 563,
            "reasoning_output_tokens": 448
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45967,
            "output_tokens": 85,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45952,
            "input_tokens": 46416,
            "output_tokens": 151,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 47107,
            "output_tokens": 572,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47488,
            "input_tokens": 47705,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 59,
        "model_tool_calls_by_name": {
          "read": 15,
          "ls": 1,
          "bash": 1,
          "exec": 27,
          "devices": 5,
          "device": 3,
          "write": 2,
          "device_stop": 4,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 389 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/aloha_bottle_grasp.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/workspace/.kinex/memos/aloha_bottle_grasp.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_cartesian.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-25-20261113/workspace/.kinex/tools/dual_cartesian.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 116,
        "observed_images": 18
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 157,
        "published_frames": 157,
        "decode_ok": true,
        "duration_s": 3.925,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "4b8c8b5f26cb09dfa46f330e298239eaffed3473e9eeea8f29c0a3bb9622d3b7",
        "source_sha256": {
          "third_person": "d3cb3da44bab586336e54951f4374dcf4cdc9c10ddbe573919b59e56226ad535",
          "left_wrist": "0d1584fb6b9564ba7f7adab24c209a48632820bda10dabc0ab31681c7706102a",
          "right_wrist": "4afe8a24f3a5d197deffa907c9296b860928de2c0f19de9ce2bb08e2ecff8197"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "393544b94431",
          "task": "robotwin/pick-diverse-bottles-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 157,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 157,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 157,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 157,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 157,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 157,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-26-20261113",
      "variant": "kinex-native",
      "task_key": "task03/26",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "26",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 7.4,
        "step_limit": 400,
        "steps": 185,
        "success": false,
        "valid": true
      },
      "steps": 185,
      "simulation_time_s": 7.4,
      "wall_time_s": 481.23271,
      "timing": {
        "simulation_time_s": 7.4,
        "wall_time_s": 481.23271,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use one arm for the bottle with red screw cap and the other for the green bottle.",
      "native_instruction": "Use one arm for the bottle with red screw cap and the other for the green bottle.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.961333571526889,
        "cache_read_input_tokens": 656512,
        "cache_reported_input_tokens": 682918,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 682918,
        "cached_input_tokens": 656512,
        "cost_usd": null,
        "fresh_input_tokens": 26406,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 682918,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 656512,
        "known_fresh_input_tokens": 26406,
        "known_input_tokens": 682918,
        "known_output_tokens": 4375,
        "known_reasoning_output_tokens": 662,
        "output_tokens": 4375,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 662,
        "reasoning_reported_output_tokens": 4375,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 30,
          "cache_write_input_tokens": 30,
          "cache_write_reported_input_tokens": 30,
          "cached_input_tokens": 30,
          "fresh_input_tokens": 30,
          "input_tokens": 30,
          "output_tokens": 30,
          "reasoning_output_tokens": 30,
          "reasoning_reported_output_tokens": 30
        },
        "request_attempts": 30,
        "response_count": 30,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 687293,
        "uncached_input_tokens": 26406,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11822,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11953,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12001,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12249,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12400,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16926,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 18189,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18759,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 18989,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19180,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 19432,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 19936,
            "output_tokens": 222,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20553,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20961,
            "output_tokens": 339,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 22920,
            "output_tokens": 238,
            "reasoning_output_tokens": 166
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23853,
            "output_tokens": 278,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24638,
            "output_tokens": 1024,
            "reasoning_output_tokens": 208
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25685,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26410,
            "output_tokens": 62,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26501,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 27089,
            "output_tokens": 191,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27309,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27904,
            "output_tokens": 132,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28687,
            "output_tokens": 70,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29384,
            "output_tokens": 121,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30142,
            "output_tokens": 183,
            "reasoning_output_tokens": 102
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 31865,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31965,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32325,
            "output_tokens": 541,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 32891,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 35,
        "model_tool_calls_by_name": {
          "read": 15,
          "ls": 1,
          "bash": 1,
          "exec": 9,
          "devices": 4,
          "device": 1,
          "write": 2,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 185 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/dual-bottle-pick.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/workspace/.kinex/memos/dual-bottle-pick.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_bottle_motion.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-26-20261113/workspace/.kinex/tools/dual_bottle_motion.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 68,
        "observed_images": 7
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 75,
        "published_frames": 75,
        "decode_ok": true,
        "duration_s": 1.875,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "d65da85bbdb618d136e719e17767204a0946c2f80653071ad9f130748f25c411",
        "source_sha256": {
          "third_person": "d18c4d7c74fab08648994f635ab0ed01fe1920c05bdceabe03a96e8d21adefae",
          "left_wrist": "8a1b13d29657ae36f6e622b8046067b7ef8f51bc4993761dc248ee38b4c4a812",
          "right_wrist": "dfdc73aaa69043838b1e6f883b40a1156a34b8a58454b1f9b6cfa614df3094a7"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "a3c0fc37880b",
          "task": "robotwin/pick-dual-bottles-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 76,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 75,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7400000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 76,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 75,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7400000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 76,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 75,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7400000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-27-20261113",
      "variant": "kinex-native",
      "task_key": "task03/27",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "27",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 8.36,
        "step_limit": 400,
        "steps": 209,
        "success": true,
        "valid": true
      },
      "steps": 209,
      "simulation_time_s": 8.36,
      "wall_time_s": 620.117267,
      "timing": {
        "simulation_time_s": 8.36,
        "wall_time_s": 620.117267,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Position the coffee-box with mug image on the left of the mouse with scroll wheel",
      "native_instruction": "Position the coffee-box with mug image on the left of the mouse with scroll wheel",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9599563656197445,
        "cache_read_input_tokens": 443520,
        "cache_reported_input_tokens": 462021,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 462021,
        "cached_input_tokens": 443520,
        "cost_usd": null,
        "fresh_input_tokens": 18501,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 462021,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 443520,
        "known_fresh_input_tokens": 18501,
        "known_input_tokens": 462021,
        "known_output_tokens": 3598,
        "known_reasoning_output_tokens": 605,
        "output_tokens": 3598,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 605,
        "reasoning_reported_output_tokens": 3598,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 23,
          "cache_write_input_tokens": 23,
          "cache_write_reported_input_tokens": 23,
          "cached_input_tokens": 23,
          "fresh_input_tokens": 23,
          "input_tokens": 23,
          "output_tokens": 23,
          "reasoning_output_tokens": 23,
          "reasoning_reported_output_tokens": 23
        },
        "request_attempts": 23,
        "response_count": 23,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 465619,
        "uncached_input_tokens": 18501,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11821,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11954,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12002,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12250,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12401,
            "output_tokens": 138,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 15294,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15232,
            "input_tokens": 16556,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 17126,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17024,
            "input_tokens": 17290,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17152,
            "input_tokens": 17481,
            "output_tokens": 200,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 19415,
            "output_tokens": 311,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20973,
            "output_tokens": 420,
            "reasoning_output_tokens": 154
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21887,
            "output_tokens": 788,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22696,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23326,
            "output_tokens": 111,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 24017,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24642,
            "output_tokens": 106,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25328,
            "output_tokens": 233,
            "reasoning_output_tokens": 140
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 26134,
            "output_tokens": 108,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26817,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27004,
            "output_tokens": 83,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27487,
            "output_tokens": 609,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28120,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 28,
        "model_tool_calls_by_name": {
          "read": 11,
          "ls": 1,
          "bash": 1,
          "exec": 10,
          "devices": 2,
          "device": 1,
          "write": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 209 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/aloha_pick_place.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/workspace/.kinex/memos/aloha_pick_place.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-27-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 54,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 85,
        "published_frames": 85,
        "decode_ok": true,
        "duration_s": 2.125,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "1fc91c43466e9e339ea94d52dea8f5895fde9341d755446e1de5305e963115ff",
        "source_sha256": {
          "third_person": "9645d6e9a8cc9b35fafa825f2aa69eda0cfab6799baac43c0c81d3a76ebabb39",
          "left_wrist": "d61e633ddfd556c0d994cf218ff0110857a87978aa4c22a4bcee0eb3b6ab97c7",
          "right_wrist": "e44d7998843eee8b10051f187b81d82f2419b1353f2d693b9ed839fefdba6f7a"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "4657936cc0c7",
          "task": "robotwin/place-a2b-left-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 85,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 85,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8360000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 85,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 85,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8360000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 85,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 85,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8360000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-28-20261113",
      "variant": "kinex-native",
      "task_key": "task03/28",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "28",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.08,
        "step_limit": 400,
        "steps": 377,
        "success": false,
        "valid": true
      },
      "steps": 377,
      "simulation_time_s": 15.08,
      "wall_time_s": 1158.755313,
      "timing": {
        "simulation_time_s": 15.08,
        "wall_time_s": 1158.755313,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Put the tiny pink car with four wheels to the right of the red, blue, green, yellow rubikscube",
      "native_instruction": "Put the tiny pink car with four wheels to the right of the red, blue, green, yellow rubikscube",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9672014225097042,
        "cache_read_input_tokens": 1177088,
        "cache_reported_input_tokens": 1217004,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1217004,
        "cached_input_tokens": 1177088,
        "cost_usd": null,
        "fresh_input_tokens": 39916,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1217004,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1177088,
        "known_fresh_input_tokens": 39916,
        "known_input_tokens": 1217004,
        "known_output_tokens": 6324,
        "known_reasoning_output_tokens": 2163,
        "output_tokens": 6324,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2163,
        "reasoning_reported_output_tokens": 6324,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 44,
          "cache_write_input_tokens": 44,
          "cache_write_reported_input_tokens": 44,
          "cached_input_tokens": 44,
          "fresh_input_tokens": 44,
          "input_tokens": 44,
          "output_tokens": 44,
          "reasoning_output_tokens": 44,
          "reasoning_reported_output_tokens": 44
        },
        "request_attempts": 44,
        "response_count": 44,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1223328,
        "uncached_input_tokens": 39916,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11831,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11866,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11999,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12049,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13517,
            "output_tokens": 172,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13568,
            "input_tokens": 16570,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 17833,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18403,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 18567,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 18758,
            "output_tokens": 184,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 20125,
            "output_tokens": 266,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 23088,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23188,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23548,
            "output_tokens": 265,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24197,
            "output_tokens": 177,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 25069,
            "output_tokens": 734,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25825,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26529,
            "output_tokens": 74,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26631,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27402,
            "output_tokens": 97,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27705,
            "output_tokens": 122,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28347,
            "output_tokens": 235,
            "reasoning_output_tokens": 161
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28775,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29368,
            "output_tokens": 111,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30003,
            "output_tokens": 145,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30928,
            "output_tokens": 61,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31780,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31882,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32393,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32998,
            "output_tokens": 233,
            "reasoning_output_tokens": 153
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33748,
            "output_tokens": 61,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33871,
            "output_tokens": 175,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34573,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34628,
            "output_tokens": 199,
            "reasoning_output_tokens": 124
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35356,
            "output_tokens": 258,
            "reasoning_output_tokens": 174
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35807,
            "output_tokens": 117,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35952,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36574,
            "output_tokens": 408,
            "reasoning_output_tokens": 313
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37511,
            "output_tokens": 134,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38173,
            "output_tokens": 456,
            "reasoning_output_tokens": 364
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39151,
            "output_tokens": 234,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39913,
            "output_tokens": 74,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40049,
            "output_tokens": 452,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40524,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 51,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 14,
          "find": 1,
          "exec": 22,
          "devices": 4,
          "device": 1,
          "write": 2,
          "edit": 2,
          "device_stop": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 377 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/tabletop.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/workspace/.kinex/memos/tabletop.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tabletop.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-28-20261113/workspace/.kinex/tools/tabletop.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 98,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 152,
        "published_frames": 152,
        "decode_ok": true,
        "duration_s": 3.8,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c303afb7588959202dd11a059e7302bcdbdda269ee9ad32df7efea070f3ede2d",
        "source_sha256": {
          "third_person": "d65586c4511e2a20384211f2ec80ffa92d8b532986ca54a619400fca20515693",
          "left_wrist": "936339bf84e2b60f0824e9219eaad0a2ed3cc2dfe492481c32e349e2f3c5c8a7",
          "right_wrist": "2a43a89aa96b8c2026c19046c6804504e65b74b884c2f9c23c19a4cf0470aa59"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "5fb1e08cd2ca",
          "task": "robotwin/place-a2b-right-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 152,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 152,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15080000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 152,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 152,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15080000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 152,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 152,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15080000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-29-20261113",
      "variant": "kinex-native",
      "task_key": "task03/29",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "29",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 15.8,
        "step_limit": 700,
        "steps": 395,
        "success": true,
        "valid": true
      },
      "steps": 395,
      "simulation_time_s": 15.8,
      "wall_time_s": 1095.118702,
      "timing": {
        "simulation_time_s": 15.8,
        "wall_time_s": 1095.118702,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move both the soft bread with golden edges and the soft rectangular bread loaf inside the white basket with beige bottom.",
      "native_instruction": "Move both the soft bread with golden edges and the soft rectangular bread loaf inside the white basket with beige bottom.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9741198748530432,
        "cache_read_input_tokens": 1115264,
        "cache_reported_input_tokens": 1144894,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1144894,
        "cached_input_tokens": 1115264,
        "cost_usd": null,
        "fresh_input_tokens": 29630,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1144894,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1115264,
        "known_fresh_input_tokens": 29630,
        "known_input_tokens": 1144894,
        "known_output_tokens": 5413,
        "known_reasoning_output_tokens": 946,
        "output_tokens": 5413,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 946,
        "reasoning_reported_output_tokens": 5413,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 40,
          "cache_write_input_tokens": 40,
          "cache_write_reported_input_tokens": 40,
          "cached_input_tokens": 40,
          "fresh_input_tokens": 40,
          "input_tokens": 40,
          "output_tokens": 40,
          "reasoning_output_tokens": 40,
          "reasoning_reported_output_tokens": 40
        },
        "request_attempts": 40,
        "response_count": 40,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1150307,
        "uncached_input_tokens": 29630,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11827,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11867,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12000,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 16356,
            "output_tokens": 43,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16256,
            "input_tokens": 16452,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 16603,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 21129,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21699,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 21929,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22120,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22372,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23305,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 24515,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24923,
            "output_tokens": 319,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25858,
            "output_tokens": 177,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 28371,
            "output_tokens": 1033,
            "reasoning_output_tokens": 146
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29425,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29723,
            "output_tokens": 164,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30421,
            "output_tokens": 45,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30528,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30633,
            "output_tokens": 163,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31223,
            "output_tokens": 214,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31955,
            "output_tokens": 111,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32128,
            "output_tokens": 128,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32743,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33032,
            "output_tokens": 117,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33643,
            "output_tokens": 335,
            "reasoning_output_tokens": 133
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34427,
            "output_tokens": 190,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34962,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35389,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35662,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36086,
            "output_tokens": 167,
            "reasoning_output_tokens": 86
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36603,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37031,
            "output_tokens": 112,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37488,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37675,
            "output_tokens": 51,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37788,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 37981,
            "output_tokens": 244,
            "reasoning_output_tokens": 144
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38252,
            "output_tokens": 495,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 38770,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 46,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 14,
          "find": 1,
          "bash": 1,
          "devices": 1,
          "device": 2,
          "exec": 19,
          "write": 2,
          "device_stop": 3,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 395 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/bread-basket.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/workspace/.kinex/memos/bread-basket.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/basket_motion.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-29-20261113/workspace/.kinex/tools/basket_motion.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 89,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 159,
        "published_frames": 159,
        "decode_ok": true,
        "duration_s": 3.975,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "5988a84c5a00827016e7f29d288e9d44644bf7c5cda709fd358eecfde8458066",
        "source_sha256": {
          "third_person": "09a98eb3ebe76aa51078e60c131c98934c8a3b872f7242d699cd9b0a702aac04",
          "left_wrist": "551799ba1df9013a69dc77e1341db52bc2b2afd8dc9236054e81859a892d9f7b",
          "right_wrist": "adf03e99f22fd1984d2544b6f84b27a84e8e094893a7eed58bc55ee65b9841b5"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "930c09a57ea5",
          "task": "robotwin/place-bread-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 160,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 159,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15800000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 160,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 159,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15800000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 160,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 159,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15800000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-30-20261113",
      "variant": "kinex-native",
      "task_key": "task03/30",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "30",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 10.84,
        "step_limit": 500,
        "steps": 271,
        "success": false,
        "valid": true
      },
      "steps": 271,
      "simulation_time_s": 10.84,
      "wall_time_s": 959.454122,
      "timing": {
        "simulation_time_s": 10.84,
        "wall_time_s": 959.454122,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move the spongy golden brown bread from the table into the versatile skillet for frying and sauteing.",
      "native_instruction": "Move the spongy golden brown bread from the table into the versatile skillet for frying and sauteing.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.965040095257981,
        "cache_read_input_tokens": 1026048,
        "cache_reported_input_tokens": 1063218,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1063218,
        "cached_input_tokens": 1026048,
        "cost_usd": null,
        "fresh_input_tokens": 37170,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1063218,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1026048,
        "known_fresh_input_tokens": 37170,
        "known_input_tokens": 1063218,
        "known_output_tokens": 7662,
        "known_reasoning_output_tokens": 2403,
        "output_tokens": 7662,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2403,
        "reasoning_reported_output_tokens": 7662,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 40,
          "cache_write_input_tokens": 40,
          "cache_write_reported_input_tokens": 40,
          "cached_input_tokens": 40,
          "fresh_input_tokens": 40,
          "input_tokens": 40,
          "output_tokens": 40,
          "reasoning_output_tokens": 40,
          "reasoning_reported_output_tokens": 40
        },
        "request_attempts": 40,
        "response_count": 40,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1070880,
        "uncached_input_tokens": 37170,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11823,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11863,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11996,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12046,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12958,
            "output_tokens": 174,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12928,
            "input_tokens": 16006,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 17282,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17152,
            "input_tokens": 17852,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18016,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18206,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18458,
            "output_tokens": 279,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 20545,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 21309,
            "output_tokens": 293,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 22190,
            "output_tokens": 265,
            "reasoning_output_tokens": 149
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 24800,
            "output_tokens": 388,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25376,
            "output_tokens": 57,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25519,
            "output_tokens": 297,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25982,
            "output_tokens": 295,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26621,
            "output_tokens": 692,
            "reasoning_output_tokens": 421
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27689,
            "output_tokens": 1020,
            "reasoning_output_tokens": 170
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28732,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29031,
            "output_tokens": 116,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29596,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30269,
            "output_tokens": 121,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30977,
            "output_tokens": 125,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31704,
            "output_tokens": 53,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31779,
            "output_tokens": 216,
            "reasoning_output_tokens": 117
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32584,
            "output_tokens": 255,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33211,
            "output_tokens": 89,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 34071,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34301,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34494,
            "output_tokens": 370,
            "reasoning_output_tokens": 266
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35450,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35643,
            "output_tokens": 534,
            "reasoning_output_tokens": 369
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36403,
            "output_tokens": 226,
            "reasoning_output_tokens": 150
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36697,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37385,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37579,
            "output_tokens": 466,
            "reasoning_output_tokens": 391
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38107,
            "output_tokens": 534,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38668,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 47,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 14,
          "find": 1,
          "exec": 18,
          "devices": 3,
          "device": 4,
          "grep": 1,
          "write": 2,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 271 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin_manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/workspace/.kinex/memos/robotwin_manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-30-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 90,
        "observed_images": 12
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 110,
        "published_frames": 110,
        "decode_ok": true,
        "duration_s": 2.75,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "ce7faa70b045c90f739ac3094f9126285d8ca5fbb80e4a3cf2f6afc551b471e7",
        "source_sha256": {
          "third_person": "81402d4d8b646814baabe7be4c1fcda70edfa09bec3346d8b85f085ed2a13859",
          "left_wrist": "7f34c22e526f9d3b3e54ace2ccbf8358e321eeda8bee92774d058e0060418855",
          "right_wrist": "ec64ddd7efc497972b334f495d425f67e19c30e064550f7bc4ad0b00ae5d3ce4"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "3aca4fb944f3",
          "task": "robotwin/place-bread-skillet-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 110,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 110,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10840000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 110,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 110,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10840000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 110,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 110,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10840000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-31-20261113",
      "variant": "kinex-native",
      "task_key": "task03/31",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "31",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.64,
        "step_limit": 500,
        "steps": 241,
        "success": true,
        "valid": true
      },
      "steps": 241,
      "simulation_time_s": 9.64,
      "wall_time_s": 881.646138,
      "timing": {
        "simulation_time_s": 9.64,
        "wall_time_s": 881.646138,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the compact hamburg box and the Crispy potato fries with red box, place them on the green tray.",
      "native_instruction": "Pick the compact hamburg box and the Crispy potato fries with red box, place them on the green tray.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9668748086673181,
        "cache_read_input_tokens": 922240,
        "cache_reported_input_tokens": 953836,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 953836,
        "cached_input_tokens": 922240,
        "cost_usd": null,
        "fresh_input_tokens": 31596,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 953836,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 922240,
        "known_fresh_input_tokens": 31596,
        "known_input_tokens": 953836,
        "known_output_tokens": 6197,
        "known_reasoning_output_tokens": 1394,
        "output_tokens": 6197,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1394,
        "reasoning_reported_output_tokens": 6197,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 32,
          "cache_write_input_tokens": 32,
          "cache_write_reported_input_tokens": 32,
          "cached_input_tokens": 32,
          "fresh_input_tokens": 32,
          "input_tokens": 32,
          "output_tokens": 32,
          "reasoning_output_tokens": 32,
          "reasoning_reported_output_tokens": 32
        },
        "request_attempts": 32,
        "response_count": 32,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 960033,
        "uncached_input_tokens": 31596,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11828,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11868,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12001,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15347,
            "output_tokens": 228,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15488,
            "input_tokens": 19255,
            "output_tokens": 148,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 21932,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22162,
            "output_tokens": 130,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 23274,
            "output_tokens": 173,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 25818,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26306,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26665,
            "output_tokens": 347,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27614,
            "output_tokens": 126,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28919,
            "output_tokens": 832,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29772,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30574,
            "output_tokens": 223,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30857,
            "output_tokens": 45,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30964,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31561,
            "output_tokens": 221,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31809,
            "output_tokens": 136,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32915,
            "output_tokens": 357,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33665,
            "output_tokens": 75,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33946,
            "output_tokens": 302,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34754,
            "output_tokens": 391,
            "reasoning_output_tokens": 166
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 36459,
            "output_tokens": 111,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 37189,
            "output_tokens": 165,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 38463,
            "output_tokens": 330,
            "reasoning_output_tokens": 205
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39710,
            "output_tokens": 330,
            "reasoning_output_tokens": 205
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40931,
            "output_tokens": 108,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41181,
            "output_tokens": 84,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41583,
            "output_tokens": 355,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 41965,
            "output_tokens": 559,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42549,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 45,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 16,
          "find": 1,
          "exec": 14,
          "devices": 5,
          "device": 1,
          "write": 2,
          "device_stop": 1,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 241 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-31-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 80,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 98,
        "published_frames": 98,
        "decode_ok": true,
        "duration_s": 2.45,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "099f43d79e82a0399e07802d492729538c0c05c954696ef51d95a0b6ae64436a",
        "source_sha256": {
          "third_person": "7a45536930bfc002a8bc61712203cde2e04ce48e7d383b4f737217b1ac1cf9d8",
          "left_wrist": "285fa36386bfd7e057abdc187a93145cad384edf1b6863394e8b08305c79162b",
          "right_wrist": "9fd607eae70439a03c76f2a3ccdc21483d22d5378cd06134d2709147fa1fed4f"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ba9f87d652c8",
          "task": "robotwin/place-burger-fries-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 98,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 98,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 98,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-32-20261113",
      "variant": "kinex-native",
      "task_key": "task03/32",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "32",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 24.84,
        "step_limit": 700,
        "steps": 621,
        "success": false,
        "valid": true
      },
      "steps": 621,
      "simulation_time_s": 24.84,
      "wall_time_s": 2529.029875,
      "timing": {
        "simulation_time_s": 24.84,
        "wall_time_s": 2529.029875,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to put the red can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "native_instruction": "Use the left arm to put the red can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.981731116600403,
        "cache_read_input_tokens": 3078912,
        "cache_reported_input_tokens": 3136207,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 3136207,
        "cached_input_tokens": 3078912,
        "cost_usd": null,
        "fresh_input_tokens": 57295,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 3136207,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 3078912,
        "known_fresh_input_tokens": 57295,
        "known_input_tokens": 3136207,
        "known_output_tokens": 13571,
        "known_reasoning_output_tokens": 5590,
        "output_tokens": 13571,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 5590,
        "reasoning_reported_output_tokens": 13571,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 74,
          "cache_write_input_tokens": 74,
          "cache_write_reported_input_tokens": 74,
          "cached_input_tokens": 74,
          "fresh_input_tokens": 74,
          "input_tokens": 74,
          "output_tokens": 74,
          "reasoning_output_tokens": 74,
          "reasoning_reported_output_tokens": 74
        },
        "request_attempts": 74,
        "response_count": 74,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 3149778,
        "uncached_input_tokens": 57295,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11829,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11869,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12002,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12052,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13900,
            "output_tokens": 205,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13952,
            "input_tokens": 17761,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 19023,
            "output_tokens": 162,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 20658,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20850,
            "output_tokens": 62,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21122,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 23024,
            "output_tokens": 114,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23724,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24132,
            "output_tokens": 377,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24531,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 26411,
            "output_tokens": 265,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 29373,
            "output_tokens": 947,
            "reasoning_output_tokens": 160
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30342,
            "output_tokens": 106,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 31066,
            "output_tokens": 72,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31166,
            "output_tokens": 104,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31957,
            "output_tokens": 60,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 33084,
            "output_tokens": 143,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33606,
            "output_tokens": 90,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33758,
            "output_tokens": 240,
            "reasoning_output_tokens": 191
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34213,
            "output_tokens": 140,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34381,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 35152,
            "output_tokens": 146,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35499,
            "output_tokens": 120,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35679,
            "output_tokens": 40,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35781,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 37015,
            "output_tokens": 147,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37607,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37970,
            "output_tokens": 647,
            "reasoning_output_tokens": 156
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38639,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39909,
            "output_tokens": 467,
            "reasoning_output_tokens": 335
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 41063,
            "output_tokens": 166,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41730,
            "output_tokens": 106,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42280,
            "output_tokens": 121,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42840,
            "output_tokens": 485,
            "reasoning_output_tokens": 355
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 44106,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44199,
            "output_tokens": 59,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 45458,
            "output_tokens": 593,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46079,
            "output_tokens": 146,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46293,
            "output_tokens": 197,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46932,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 48439,
            "output_tokens": 57,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48128,
            "input_tokens": 48820,
            "output_tokens": 249,
            "reasoning_output_tokens": 139
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49508,
            "output_tokens": 57,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 49627,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 50152,
            "output_tokens": 180,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50176,
            "input_tokens": 50360,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50304,
            "input_tokens": 52019,
            "output_tokens": 771,
            "reasoning_output_tokens": 650
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52608,
            "input_tokens": 53222,
            "output_tokens": 320,
            "reasoning_output_tokens": 209
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53376,
            "input_tokens": 53982,
            "output_tokens": 60,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53888,
            "input_tokens": 54104,
            "output_tokens": 118,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54016,
            "input_tokens": 54547,
            "output_tokens": 204,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54528,
            "input_tokens": 56420,
            "output_tokens": 687,
            "reasoning_output_tokens": 298
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56960,
            "input_tokens": 57580,
            "output_tokens": 377,
            "reasoning_output_tokens": 272
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57728,
            "input_tokens": 58390,
            "output_tokens": 228,
            "reasoning_output_tokens": 130
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58368,
            "input_tokens": 59061,
            "output_tokens": 157,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59008,
            "input_tokens": 59596,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59392,
            "input_tokens": 59788,
            "output_tokens": 134,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59904,
            "input_tokens": 60360,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60288,
            "input_tokens": 60550,
            "output_tokens": 97,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60544,
            "input_tokens": 62098,
            "output_tokens": 392,
            "reasoning_output_tokens": 286
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62336,
            "input_tokens": 62936,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62848,
            "input_tokens": 64169,
            "output_tokens": 101,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 64128,
            "input_tokens": 64594,
            "output_tokens": 473,
            "reasoning_output_tokens": 370
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 64896,
            "input_tokens": 65504,
            "output_tokens": 46,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 65408,
            "input_tokens": 65612,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 65536,
            "input_tokens": 65975,
            "output_tokens": 248,
            "reasoning_output_tokens": 127
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 66048,
            "input_tokens": 66665,
            "output_tokens": 59,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 66560,
            "input_tokens": 66895,
            "output_tokens": 217,
            "reasoning_output_tokens": 177
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 66944,
            "input_tokens": 67174,
            "output_tokens": 795,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 67840,
            "input_tokens": 67995,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 83,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "find": 1,
          "exec": 37,
          "devices": 8,
          "device": 4,
          "write": 4,
          "edit": 4,
          "device_stop": 7
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 621 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/aloha_basket.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/workspace/.kinex/memos/aloha_basket.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cloud_view.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/workspace/.kinex/tools/cloud_view.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/workspace/.kinex/tools/move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/observe.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-32-20261113/workspace/.kinex/tools/observe.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 160,
        "observed_images": 28
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 250,
        "published_frames": 250,
        "decode_ok": true,
        "duration_s": 6.25,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "8ba0a93304efa7378d63d0110a80ac26cafaf48d3b64040382f135f8c1d9746a",
        "source_sha256": {
          "third_person": "09ab67c0ce1ae1db6695e0b307186877d6016f7b88ad03b1483933299352e085",
          "left_wrist": "c70d69c4e2b54bf20b7a807f40d17a18fa14c15bc4d3205b41a0f79f8cfa39bb",
          "right_wrist": "ead5ba136055071b4d7e99eb8a70817bd9e37bb01db032c713d27a1ca6fb4f5e"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ce616f565a56",
          "task": "robotwin/place-can-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 250,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 250,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24840000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 250,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 250,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24840000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 250,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 250,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24840000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-33-20261113",
      "variant": "kinex-native",
      "task_key": "task03/33",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "33",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 17.88,
        "step_limit": 800,
        "steps": 447,
        "success": true,
        "valid": true
      },
      "steps": 447,
      "simulation_time_s": 17.88,
      "wall_time_s": 1381.201167,
      "timing": {
        "simulation_time_s": 17.88,
        "wall_time_s": 1381.201167,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the small red soda can and the lightweight cylindrical metal can in parallel, then put both into the rectangular plasticbox.",
      "native_instruction": "Grab the small red soda can and the lightweight cylindrical metal can in parallel, then put both into the rectangular plasticbox.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9734321511193614,
        "cache_read_input_tokens": 1880448,
        "cache_reported_input_tokens": 1931771,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1931771,
        "cached_input_tokens": 1880448,
        "cost_usd": null,
        "fresh_input_tokens": 51323,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1931771,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1880448,
        "known_fresh_input_tokens": 51323,
        "known_input_tokens": 1931771,
        "known_output_tokens": 7656,
        "known_reasoning_output_tokens": 1959,
        "output_tokens": 7656,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1959,
        "reasoning_reported_output_tokens": 7656,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 44,
          "cache_write_input_tokens": 44,
          "cache_write_reported_input_tokens": 44,
          "cached_input_tokens": 44,
          "fresh_input_tokens": 44,
          "input_tokens": 44,
          "output_tokens": 44,
          "reasoning_output_tokens": 44,
          "reasoning_reported_output_tokens": 44
        },
        "request_attempts": 44,
        "response_count": 44,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1939427,
        "uncached_input_tokens": 51323,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11827,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11867,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12000,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12050,
            "output_tokens": 145,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12403,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16929,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 18191,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 18421,
            "output_tokens": 107,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 19918,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 20170,
            "output_tokens": 183,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 41392,
            "output_tokens": 338,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42910,
            "output_tokens": 133,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43509,
            "output_tokens": 272,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 43981,
            "output_tokens": 110,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 45270,
            "output_tokens": 1070,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46361,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 47079,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47156,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 47526,
            "output_tokens": 168,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47488,
            "input_tokens": 47721,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 48539,
            "output_tokens": 360,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48768,
            "input_tokens": 49149,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49288,
            "output_tokens": 39,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49152,
            "input_tokens": 49389,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49280,
            "input_tokens": 49986,
            "output_tokens": 124,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49920,
            "input_tokens": 50860,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50688,
            "input_tokens": 51413,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51456,
            "input_tokens": 52241,
            "output_tokens": 262,
            "reasoning_output_tokens": 191
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52352,
            "input_tokens": 53121,
            "output_tokens": 248,
            "reasoning_output_tokens": 178
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53248,
            "input_tokens": 53982,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53888,
            "input_tokens": 54531,
            "output_tokens": 150,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54528,
            "input_tokens": 55292,
            "output_tokens": 233,
            "reasoning_output_tokens": 147
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55296,
            "input_tokens": 56263,
            "output_tokens": 252,
            "reasoning_output_tokens": 183
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56320,
            "input_tokens": 57128,
            "output_tokens": 82,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56960,
            "input_tokens": 57232,
            "output_tokens": 247,
            "reasoning_output_tokens": 162
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57344,
            "input_tokens": 57506,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57472,
            "input_tokens": 57779,
            "output_tokens": 327,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57984,
            "input_tokens": 58476,
            "output_tokens": 211,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58624,
            "input_tokens": 59304,
            "output_tokens": 101,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59264,
            "input_tokens": 60020,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60032,
            "input_tokens": 60221,
            "output_tokens": 76,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60160,
            "input_tokens": 60557,
            "output_tokens": 459,
            "reasoning_output_tokens": 173
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60928,
            "input_tokens": 61043,
            "output_tokens": 703,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61568,
            "input_tokens": 61770,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 54,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 13,
          "find": 4,
          "exec": 23,
          "devices": 3,
          "device": 1,
          "write": 2,
          "edit": 4,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 447 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-cans.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/workspace/.kinex/memos/robotwin-cans.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_tcp.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-33-20261113/workspace/.kinex/tools/dual_tcp.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 101,
        "observed_images": 15
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 180,
        "published_frames": 180,
        "decode_ok": true,
        "duration_s": 4.5,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "dc29e5b900623c501674e8a2829c0a8808fe078dfd7ee75ac9d3368d038af9c9",
        "source_sha256": {
          "third_person": "e69f719e24fd09a1cd992041e75479c7d9acf476ae8f312963de53b7d972d6ae",
          "left_wrist": "a8d41fd40116e72f7e3d1c27535ad6a69b492f412d9ccf7619fb96bc66113252",
          "right_wrist": "71f64b3de0d8ad72e9b99fa5960f6e714e7d49d437e418e9dec24a38be9a7fd1"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "e777e063ed91",
          "task": "robotwin/place-cans-plasticbox-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 180,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 180,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17880000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 180,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 180,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17880000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 180,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 180,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17880000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-34-20261113",
      "variant": "kinex-native",
      "task_key": "task03/34",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "34",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 8.72,
        "step_limit": 400,
        "steps": 218,
        "success": true,
        "valid": true
      },
      "steps": 218,
      "simulation_time_s": 8.72,
      "wall_time_s": 836.797927,
      "timing": {
        "simulation_time_s": 8.72,
        "wall_time_s": 836.797927,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the ceramic drinking cup with floral design on the gray plate for dining.",
      "native_instruction": "Place the ceramic drinking cup with floral design on the gray plate for dining.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.934713147980352,
        "cache_read_input_tokens": 940800,
        "cache_reported_input_tokens": 1006512,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1006512,
        "cached_input_tokens": 940800,
        "cost_usd": null,
        "fresh_input_tokens": 65712,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1006512,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 940800,
        "known_fresh_input_tokens": 65712,
        "known_input_tokens": 1006512,
        "known_output_tokens": 5314,
        "known_reasoning_output_tokens": 791,
        "output_tokens": 5314,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 791,
        "reasoning_reported_output_tokens": 5314,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 35,
          "cache_write_input_tokens": 35,
          "cache_write_reported_input_tokens": 35,
          "cached_input_tokens": 35,
          "fresh_input_tokens": 35,
          "input_tokens": 35,
          "output_tokens": 35,
          "reasoning_output_tokens": 35,
          "reasoning_reported_output_tokens": 35
        },
        "request_attempts": 35,
        "response_count": 35,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1011826,
        "uncached_input_tokens": 65712,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11991,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14713,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14592,
            "input_tokens": 14987,
            "output_tokens": 162,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14976,
            "input_tokens": 17359,
            "output_tokens": 188,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 22889,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23119,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 24728,
            "output_tokens": 61,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25158,
            "output_tokens": 332,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26295,
            "output_tokens": 432,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27347,
            "output_tokens": 163,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 29382,
            "output_tokens": 1115,
            "reasoning_output_tokens": 196
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30518,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 31233,
            "output_tokens": 85,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31345,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31611,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31755,
            "output_tokens": 53,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31870,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32356,
            "output_tokens": 78,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 32466,
            "output_tokens": 130,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32623,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33057,
            "output_tokens": 121,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33532,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33724,
            "output_tokens": 162,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 34215,
            "output_tokens": 139,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34704,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34895,
            "output_tokens": 127,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35371,
            "output_tokens": 190,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35909,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36100,
            "output_tokens": 152,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36394,
            "output_tokens": 73,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36729,
            "output_tokens": 226,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 36982,
            "output_tokens": 472,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37479,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 46,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 14,
          "find": 1,
          "bash": 2,
          "exec": 14,
          "devices": 3,
          "device": 4,
          "write": 2,
          "edit": 3,
          "device_stop": 1,
          "grep": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 218 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-34-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 84,
        "observed_images": 6
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 88,
        "published_frames": 88,
        "decode_ok": true,
        "duration_s": 2.2,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "2ee0324e67871b9bf07c7e6e0ff16b5ee76afeb6df6eb319b1d7db4de3f86f0d",
        "source_sha256": {
          "third_person": "fa1aae8184e3c2c857ec93e3941d8e199252a8de2bc6057a411d068e1e494cf9",
          "left_wrist": "71b713736375f95317363b132356c1ad51381cd52434a6ba8051e0a4e7fc6d0a",
          "right_wrist": "b3df508d23fefb8782ad435e46b92ef74f84263290c782f7dbaae019bb08697a"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "67e82452d973",
          "task": "robotwin/place-container-plate-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 89,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 88,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8720000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 89,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 88,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8720000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 89,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 88,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8720000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-35-20261113",
      "variant": "kinex-native",
      "task_key": "task03/35",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "35",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 10.08,
        "step_limit": 500,
        "steps": 252,
        "success": true,
        "valid": true
      },
      "steps": 252,
      "simulation_time_s": 10.08,
      "wall_time_s": 867.298659,
      "timing": {
        "simulation_time_s": 10.08,
        "wall_time_s": 867.298659,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the small smooth cup onto the light gray coaster gently.",
      "native_instruction": "Place the small smooth cup onto the light gray coaster gently.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9652693671815843,
        "cache_read_input_tokens": 935680,
        "cache_reported_input_tokens": 969346,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 969346,
        "cached_input_tokens": 935680,
        "cost_usd": null,
        "fresh_input_tokens": 33666,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 969346,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 935680,
        "known_fresh_input_tokens": 33666,
        "known_input_tokens": 969346,
        "known_output_tokens": 5695,
        "known_reasoning_output_tokens": 913,
        "output_tokens": 5695,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 913,
        "reasoning_reported_output_tokens": 5695,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 43,
          "cache_write_input_tokens": 43,
          "cache_write_reported_input_tokens": 43,
          "cached_input_tokens": 43,
          "fresh_input_tokens": 43,
          "input_tokens": 43,
          "output_tokens": 43,
          "reasoning_output_tokens": 43,
          "reasoning_reported_output_tokens": 43
        },
        "request_attempts": 43,
        "response_count": 43,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 975041,
        "uncached_input_tokens": 33666,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11815,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11850,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11983,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12031,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12279,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12742,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12672,
            "input_tokens": 13556,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13440,
            "input_tokens": 14373,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14208,
            "input_tokens": 15199,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15104,
            "input_tokens": 15991,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 16142,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 17404,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 17634,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 17825,
            "output_tokens": 69,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18425,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 18677,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 19456,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 19864,
            "output_tokens": 336,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20096,
            "input_tokens": 20735,
            "output_tokens": 321,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21634,
            "output_tokens": 139,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 22239,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 23458,
            "output_tokens": 429,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24306,
            "output_tokens": 66,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25067,
            "output_tokens": 1069,
            "reasoning_output_tokens": 153
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26157,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26387,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26533,
            "output_tokens": 47,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26642,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27061,
            "output_tokens": 218,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27306,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27806,
            "output_tokens": 133,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28369,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28564,
            "output_tokens": 78,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28954,
            "output_tokens": 125,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29518,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29712,
            "output_tokens": 221,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30427,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30621,
            "output_tokens": 317,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31149,
            "output_tokens": 194,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31864,
            "output_tokens": 113,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32119,
            "output_tokens": 63,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32410,
            "output_tokens": 628,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33062,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 42,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 13,
          "bash": 1,
          "exec": 15,
          "devices": 3,
          "device": 4,
          "write": 2,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 252 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-cup.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/workspace/.kinex/memos/robotwin-cup.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-35-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 88,
        "observed_images": 6
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 102,
        "published_frames": 102,
        "decode_ok": true,
        "duration_s": 2.55,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "3fc5da194fa1e0d0b4187a7b9c8d3f6abdd03ca3b489a7ebab8dcb312263303b",
        "source_sha256": {
          "third_person": "43b98455a99be9c20a534497e28441f35e4234887ada9b01c83d5d7b1f4a4ac2",
          "left_wrist": "d2ae778be42c2cd95f8f6ac7133f4d9c0f270d607b8eda8c06b4fdef935a994e",
          "right_wrist": "81369bdaa2483b0e936240554aaad2e048fb0da1e01edbc6e8caad4e971293ff"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "a4fc6ed7e890",
          "task": "robotwin/place-empty-cup-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 102,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 102,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10080000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 102,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 102,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10080000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 102,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 102,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10080000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-36-20261113",
      "variant": "kinex-native",
      "task_key": "task03/36",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "36",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 8.56,
        "step_limit": 400,
        "steps": 214,
        "success": true,
        "valid": true
      },
      "steps": 214,
      "simulation_time_s": 8.56,
      "wall_time_s": 1047.7674,
      "timing": {
        "simulation_time_s": 8.56,
        "wall_time_s": 1047.7674,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the beige circular blade fan, place it on the Pink mat, face it toward the robot",
      "native_instruction": "Pick the beige circular blade fan, place it on the Pink mat, face it toward the robot",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9726220775733186,
        "cache_read_input_tokens": 1213312,
        "cache_reported_input_tokens": 1247465,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1247465,
        "cached_input_tokens": 1213312,
        "cost_usd": null,
        "fresh_input_tokens": 34153,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1247465,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1213312,
        "known_fresh_input_tokens": 34153,
        "known_input_tokens": 1247465,
        "known_output_tokens": 7487,
        "known_reasoning_output_tokens": 2227,
        "output_tokens": 7487,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2227,
        "reasoning_reported_output_tokens": 7487,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 42,
          "cache_write_input_tokens": 42,
          "cache_write_reported_input_tokens": 42,
          "cached_input_tokens": 42,
          "fresh_input_tokens": 42,
          "input_tokens": 42,
          "output_tokens": 42,
          "reasoning_output_tokens": 42,
          "reasoning_reported_output_tokens": 42
        },
        "request_attempts": 42,
        "response_count": 42,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1254952,
        "uncached_input_tokens": 34153,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11826,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11866,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11999,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12049,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14525,
            "output_tokens": 204,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14592,
            "input_tokens": 18388,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 19650,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 19880,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 20070,
            "output_tokens": 105,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20706,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 20958,
            "output_tokens": 152,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 23876,
            "output_tokens": 289,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24560,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24968,
            "output_tokens": 289,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25792,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 27637,
            "output_tokens": 306,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28397,
            "output_tokens": 325,
            "reasoning_output_tokens": 251
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29901,
            "output_tokens": 188,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30300,
            "output_tokens": 911,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31234,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31514,
            "output_tokens": 124,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32067,
            "output_tokens": 321,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32762,
            "output_tokens": 211,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33403,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33881,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34404,
            "output_tokens": 605,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35069,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35123,
            "output_tokens": 62,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35554,
            "output_tokens": 333,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36234,
            "output_tokens": 491,
            "reasoning_output_tokens": 258
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 36901,
            "output_tokens": 197,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37961,
            "output_tokens": 326,
            "reasoning_output_tokens": 160
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 39053,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39142,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39450,
            "output_tokens": 79,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39989,
            "output_tokens": 226,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40244,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40382,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40912,
            "output_tokens": 138,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41192,
            "output_tokens": 87,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41540,
            "output_tokens": 539,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42106,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 49,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "find": 1,
          "exec": 19,
          "devices": 4,
          "device": 1,
          "write": 2,
          "device_stop": 2,
          "edit": 1,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 214 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-36-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 94,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 87,
        "published_frames": 87,
        "decode_ok": true,
        "duration_s": 2.175,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "622eff873bc3aa2004592f31fcf80633d9d76e3f7b535ab09f7a6731cbac0e87",
        "source_sha256": {
          "third_person": "89b8a281c5c604bb6844903fa82d23b921c2fbc3a3cec4345f10f23d8c3673e5",
          "left_wrist": "3cc084c14220ae7f3feca80b1bd68bd2f1df7d2591ec7667abde9d1381ec4e4f",
          "right_wrist": "69147cae4479cedabb3b2cc5c208182b2da381f6770b0e33a602bc91586d76f6"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "e5ae610b9d91",
          "task": "robotwin/place-fan-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 87,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 87,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 87,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 87,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 87,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 87,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-37-20261113",
      "variant": "kinex-native",
      "task_key": "task03/37",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "37",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 2
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 8.2,
        "step_limit": 400,
        "steps": 205,
        "success": true,
        "valid": true
      },
      "steps": 205,
      "simulation_time_s": 8.2,
      "wall_time_s": 762.046817,
      "timing": {
        "simulation_time_s": 8.2,
        "wall_time_s": 762.046817,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the dark gray computer mouse with logo with the left arm and drop it onto the Cyan mat.",
      "native_instruction": "Pick up the dark gray computer mouse with logo with the left arm and drop it onto the Cyan mat.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9706699022330074,
        "cache_read_input_tokens": 665600,
        "cache_reported_input_tokens": 685712,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 685712,
        "cached_input_tokens": 665600,
        "cost_usd": null,
        "fresh_input_tokens": 20112,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 685712,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 665600,
        "known_fresh_input_tokens": 20112,
        "known_input_tokens": 685712,
        "known_output_tokens": 4073,
        "known_reasoning_output_tokens": 546,
        "output_tokens": 4073,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 546,
        "reasoning_reported_output_tokens": 4073,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 32,
          "cache_write_input_tokens": 32,
          "cache_write_reported_input_tokens": 32,
          "cached_input_tokens": 32,
          "fresh_input_tokens": 32,
          "input_tokens": 32,
          "output_tokens": 32,
          "reasoning_output_tokens": 32,
          "reasoning_reported_output_tokens": 32
        },
        "request_attempts": 32,
        "response_count": 32,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 689785,
        "uncached_input_tokens": 20112,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11821,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11952,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12000,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12248,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 13062,
            "output_tokens": 134,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13056,
            "input_tokens": 15292,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15232,
            "input_tokens": 16554,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 16718,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 18147,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18399,
            "output_tokens": 276,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 19070,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 19478,
            "output_tokens": 276,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20332,
            "output_tokens": 121,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 21612,
            "output_tokens": 884,
            "reasoning_output_tokens": 211
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22518,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 23221,
            "output_tokens": 70,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23319,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23458,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23513,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23932,
            "output_tokens": 179,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24139,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24630,
            "output_tokens": 98,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25169,
            "output_tokens": 126,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25705,
            "output_tokens": 55,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25782,
            "output_tokens": 84,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26230,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26760,
            "output_tokens": 134,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27362,
            "output_tokens": 92,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27890,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28085,
            "output_tokens": 56,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28370,
            "output_tokens": 549,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 28944,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 37,
        "model_tool_calls_by_name": {
          "read": 11,
          "ls": 1,
          "bash": 1,
          "exec": 14,
          "devices": 3,
          "device": 1,
          "write": 2,
          "edit": 2,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 205 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/aloha_mouse_placement.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/workspace/.kinex/memos/aloha_mouse_placement.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/move_left_tcp.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-37-20261113/workspace/.kinex/tools/move_left_tcp.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 72,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 83,
        "published_frames": 83,
        "decode_ok": true,
        "duration_s": 2.075,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "77f97eeae5ed3ebdca7c0a6a7bc6031723884e505fa663d5e2590f11f33006e9",
        "source_sha256": {
          "third_person": "361a76611429f264238c26128cc0bc8fb889f6776ac598f510c0b6c2739fb363",
          "left_wrist": "12d5b130404dc05c1e9e931de776a25b0e7e60c24f7341374eb468b4f26db531",
          "right_wrist": "01c3d68277bf0218743f0c3a8b4b4b2b4bfbf02def4f770e89894744f262a436"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "db4ddb9f1d6c",
          "task": "robotwin/place-mouse-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 84,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 83,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 84,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 83,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 84,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 83,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-38-20261113",
      "variant": "kinex-native",
      "task_key": "task03/38",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "38",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 18.96,
        "step_limit": 700,
        "steps": 474,
        "success": false,
        "valid": true
      },
      "steps": 474,
      "simulation_time_s": 18.96,
      "wall_time_s": 1693.433726,
      "timing": {
        "simulation_time_s": 18.96,
        "wall_time_s": 1693.433726,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the toycar with rounded edges and place into the basket with rounded edges.",
      "native_instruction": "Grab the toycar with rounded edges and place into the basket with rounded edges.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9743839757072613,
        "cache_read_input_tokens": 1399040,
        "cache_reported_input_tokens": 1435820,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1435820,
        "cached_input_tokens": 1399040,
        "cost_usd": null,
        "fresh_input_tokens": 36780,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1435820,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1399040,
        "known_fresh_input_tokens": 36780,
        "known_input_tokens": 1435820,
        "known_output_tokens": 10613,
        "known_reasoning_output_tokens": 3735,
        "output_tokens": 10613,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3735,
        "reasoning_reported_output_tokens": 10613,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 45,
          "cache_write_input_tokens": 45,
          "cache_write_reported_input_tokens": 45,
          "cached_input_tokens": 45,
          "fresh_input_tokens": 45,
          "input_tokens": 45,
          "output_tokens": 45,
          "reasoning_output_tokens": 45,
          "reasoning_reported_output_tokens": 45
        },
        "request_attempts": 45,
        "response_count": 45,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1446433,
        "uncached_input_tokens": 36780,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11822,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11862,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11995,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12045,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15182,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15104,
            "input_tokens": 15430,
            "output_tokens": 205,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15488,
            "input_tokens": 19291,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 20554,
            "output_tokens": 166,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 22008,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22199,
            "output_tokens": 160,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 23400,
            "output_tokens": 318,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 25209,
            "output_tokens": 311,
            "reasoning_output_tokens": 233
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26699,
            "output_tokens": 922,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27645,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28105,
            "output_tokens": 171,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28334,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28400,
            "output_tokens": 103,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28917,
            "output_tokens": 203,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29150,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29914,
            "output_tokens": 271,
            "reasoning_output_tokens": 92
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30697,
            "output_tokens": 221,
            "reasoning_output_tokens": 182
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 30940,
            "output_tokens": 454,
            "reasoning_output_tokens": 169
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31805,
            "output_tokens": 107,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32883,
            "output_tokens": 320,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33681,
            "output_tokens": 359,
            "reasoning_output_tokens": 209
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34070,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34572,
            "output_tokens": 140,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35464,
            "output_tokens": 181,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36413,
            "output_tokens": 202,
            "reasoning_output_tokens": 102
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37251,
            "output_tokens": 72,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37345,
            "output_tokens": 154,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 38137,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38766,
            "output_tokens": 182,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39617,
            "output_tokens": 343,
            "reasoning_output_tokens": 217
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40682,
            "output_tokens": 733,
            "reasoning_output_tokens": 205
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 42099,
            "output_tokens": 618,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 43409,
            "output_tokens": 178,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 44051,
            "output_tokens": 374,
            "reasoning_output_tokens": 148
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 45456,
            "output_tokens": 737,
            "reasoning_output_tokens": 432
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46347,
            "output_tokens": 335,
            "reasoning_output_tokens": 206
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 47349,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47360,
            "input_tokens": 47973,
            "output_tokens": 265,
            "reasoning_output_tokens": 162
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48128,
            "input_tokens": 48933,
            "output_tokens": 603,
            "reasoning_output_tokens": 199
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49408,
            "input_tokens": 49566,
            "output_tokens": 560,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50048,
            "input_tokens": 50153,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 58,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 14,
          "find": 1,
          "bash": 1,
          "exec": 24,
          "device": 8,
          "write": 2,
          "device_stop": 3,
          "edit": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 474 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/aloha_manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/workspace/.kinex/memos/aloha_manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/aloha_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-38-20261113/workspace/.kinex/tools/aloha_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 106,
        "observed_images": 15
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 191,
        "published_frames": 191,
        "decode_ok": true,
        "duration_s": 4.775,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "0bfe301e8c8138dc8903b94d832e41c543665e7f388a97d52dc9fdefb96b73a7",
        "source_sha256": {
          "third_person": "83aee4df6926fe178b84102bda4479fb7b9c5bd85f4e4f477fd473ac45e1a649",
          "left_wrist": "75a4e6ebf4d5a41729662706e0479566a2e8b2a7b8823f928afb33adf40335da",
          "right_wrist": "1c196d73d5eb91e8e523fababdb5a69ff0da3ba0cd4fd89474decfd535628bdd"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "08f27c85b634",
          "task": "robotwin/place-object-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 191,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 191,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 191,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 191,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 191,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 191,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-39-20261113",
      "variant": "kinex-native",
      "task_key": "task03/39",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "39",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 8.72,
        "step_limit": 400,
        "steps": 218,
        "success": true,
        "valid": true
      },
      "steps": 218,
      "simulation_time_s": 8.72,
      "wall_time_s": 847.187583,
      "timing": {
        "simulation_time_s": 8.72,
        "wall_time_s": 847.187583,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use one arm to pick the palm-sized black stapler for the black screen electronicscale",
      "native_instruction": "Use one arm to pick the palm-sized black stapler for the black screen electronicscale",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.972222253797459,
        "cache_read_input_tokens": 855296,
        "cache_reported_input_tokens": 879733,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 879733,
        "cached_input_tokens": 855296,
        "cost_usd": null,
        "fresh_input_tokens": 24437,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 879733,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 855296,
        "known_fresh_input_tokens": 24437,
        "known_input_tokens": 879733,
        "known_output_tokens": 4448,
        "known_reasoning_output_tokens": 638,
        "output_tokens": 4448,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 638,
        "reasoning_reported_output_tokens": 4448,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 34,
          "cache_write_input_tokens": 34,
          "cache_write_reported_input_tokens": 34,
          "cached_input_tokens": 34,
          "fresh_input_tokens": 34,
          "input_tokens": 34,
          "output_tokens": 34,
          "reasoning_output_tokens": 34,
          "reasoning_reported_output_tokens": 34
        },
        "request_attempts": 34,
        "response_count": 34,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 884181,
        "uncached_input_tokens": 24437,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11991,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15172,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15104,
            "input_tokens": 15443,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15488,
            "input_tokens": 19292,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 20554,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20718,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 22149,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22401,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22905,
            "output_tokens": 346,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23646,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24054,
            "output_tokens": 298,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25135,
            "output_tokens": 185,
            "reasoning_output_tokens": 143
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 26499,
            "output_tokens": 807,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27327,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27462,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28069,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28676,
            "output_tokens": 72,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29263,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30121,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30177,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30666,
            "output_tokens": 224,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 30917,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31070,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31127,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31545,
            "output_tokens": 117,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31768,
            "output_tokens": 166,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 31961,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32638,
            "output_tokens": 81,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32861,
            "output_tokens": 57,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33146,
            "output_tokens": 254,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33427,
            "output_tokens": 425,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33877,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 40,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 12,
          "find": 1,
          "bash": 1,
          "exec": 12,
          "devices": 4,
          "device": 1,
          "write": 2,
          "device_stop": 2,
          "edit": 3,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 218 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-39-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 77,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 88,
        "published_frames": 88,
        "decode_ok": true,
        "duration_s": 2.2,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "ed5f29329e523f2ef60346d28581a61b2db7d93dce5d9c78bbf5a704bb81b506",
        "source_sha256": {
          "third_person": "3e6a6a84cc9bc9125aaba440b3fbcdb715fc27eb66c6bf355920d6252d1cdbec",
          "left_wrist": "67b886856939ff91bf8a4e895af76388b48faef0452fd228f2b5f4164c1b4831",
          "right_wrist": "ee38954865c6bd9fb6cd3a3ea610b4e51f1b5e69e89a00df168a6015f0cd02d4"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "7c0c26ae07c0",
          "task": "robotwin/place-object-scale-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 89,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 88,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8720000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 89,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 88,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8720000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 89,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 88,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8720000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-40-20261113",
      "variant": "kinex-native",
      "task_key": "task03/40",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "40",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 11.56,
        "step_limit": 400,
        "steps": 289,
        "success": true,
        "valid": true
      },
      "steps": 289,
      "simulation_time_s": 11.56,
      "wall_time_s": 915.602627,
      "timing": {
        "simulation_time_s": 11.56,
        "wall_time_s": 915.602627,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set the rectangular remotecontrol on the displaystand with hollow center.",
      "native_instruction": "Set the rectangular remotecontrol on the displaystand with hollow center.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9713396581683004,
        "cache_read_input_tokens": 802176,
        "cache_reported_input_tokens": 825845,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 825845,
        "cached_input_tokens": 802176,
        "cost_usd": null,
        "fresh_input_tokens": 23669,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 825845,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 802176,
        "known_fresh_input_tokens": 23669,
        "known_input_tokens": 825845,
        "known_output_tokens": 5642,
        "known_reasoning_output_tokens": 1078,
        "output_tokens": 5642,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1078,
        "reasoning_reported_output_tokens": 5642,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 831487,
        "uncached_input_tokens": 23669,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11946,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11994,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12242,
            "output_tokens": 223,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 16913,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16768,
            "input_tokens": 18497,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 18727,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 18917,
            "output_tokens": 77,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19525,
            "output_tokens": 180,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 21352,
            "output_tokens": 337,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 22224,
            "output_tokens": 330,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 23104,
            "output_tokens": 513,
            "reasoning_output_tokens": 169
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23917,
            "output_tokens": 124,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 25220,
            "output_tokens": 1030,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26271,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26522,
            "output_tokens": 120,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27095,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27616,
            "output_tokens": 243,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28140,
            "output_tokens": 65,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28666,
            "output_tokens": 269,
            "reasoning_output_tokens": 202
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 28993,
            "output_tokens": 64,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29079,
            "output_tokens": 76,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29526,
            "output_tokens": 221,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29774,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30615,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30672,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31087,
            "output_tokens": 161,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31275,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31887,
            "output_tokens": 128,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32527,
            "output_tokens": 158,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32827,
            "output_tokens": 62,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33118,
            "output_tokens": 619,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33762,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 41,
        "model_tool_calls_by_name": {
          "read": 13,
          "ls": 1,
          "bash": 1,
          "exec": 15,
          "devices": 4,
          "device": 1,
          "write": 2,
          "device_stop": 2,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 289 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-arx5.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/workspace/.kinex/memos/robotwin-arx5.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-40-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 77,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 117,
        "published_frames": 117,
        "decode_ok": true,
        "duration_s": 2.925,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "d5e8a66321e093b889ad3bf236d514c7b56b842bec2c38bff296b5244256d522",
        "source_sha256": {
          "third_person": "464fe299671bf1dd30ff0159a6c877f1c5f3a152ee531c321346604fa6c1cfb8",
          "left_wrist": "9cfce00b22fdccebd8be2816948e4fa38e80f7c21806870f4ce04341fabe04a4",
          "right_wrist": "d7e97ffbff4e3611264fb0fd5cd4fce72d95597a0bac0264cf99d0af65049008"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "f2574403dcf3",
          "task": "robotwin/place-object-stand-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 117,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 117,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 117,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 117,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 117,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 117,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-41-20261113",
      "variant": "kinex-native",
      "task_key": "task03/41",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "41",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.84,
        "step_limit": 400,
        "steps": 246,
        "success": true,
        "valid": true
      },
      "steps": 246,
      "simulation_time_s": 9.84,
      "wall_time_s": 826.53522,
      "timing": {
        "simulation_time_s": 9.84,
        "wall_time_s": 826.53522,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set the phone with shiny black back on the L-shaped smooth brown phonestand.",
      "native_instruction": "Set the phone with shiny black back on the L-shaped smooth brown phonestand.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9587300289624378,
        "cache_read_input_tokens": 797440,
        "cache_reported_input_tokens": 831767,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 831767,
        "cached_input_tokens": 797440,
        "cost_usd": null,
        "fresh_input_tokens": 34327,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 831767,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 797440,
        "known_fresh_input_tokens": 34327,
        "known_input_tokens": 831767,
        "known_output_tokens": 5846,
        "known_reasoning_output_tokens": 1877,
        "output_tokens": 5846,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1877,
        "reasoning_reported_output_tokens": 5846,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 34,
          "cache_write_input_tokens": 34,
          "cache_write_reported_input_tokens": 34,
          "cached_input_tokens": 34,
          "fresh_input_tokens": 34,
          "input_tokens": 34,
          "output_tokens": 34,
          "reasoning_output_tokens": 34,
          "reasoning_reported_output_tokens": 34
        },
        "request_attempts": 34,
        "response_count": 34,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 837613,
        "uncached_input_tokens": 34327,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11859,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11992,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12291,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12754,
            "output_tokens": 165,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 16146,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12800,
            "input_tokens": 17408,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 17638,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 17827,
            "output_tokens": 150,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 19966,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 20218,
            "output_tokens": 317,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20930,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21338,
            "output_tokens": 315,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 23049,
            "output_tokens": 88,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23603,
            "output_tokens": 780,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24404,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24558,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25214,
            "output_tokens": 109,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25880,
            "output_tokens": 68,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26466,
            "output_tokens": 135,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27175,
            "output_tokens": 409,
            "reasoning_output_tokens": 124
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 29429,
            "output_tokens": 524,
            "reasoning_output_tokens": 425
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30018,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30213,
            "output_tokens": 51,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30286,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30704,
            "output_tokens": 91,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30901,
            "output_tokens": 138,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31614,
            "output_tokens": 62,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32447,
            "output_tokens": 330,
            "reasoning_output_tokens": 234
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33347,
            "output_tokens": 715,
            "reasoning_output_tokens": 619
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34634,
            "output_tokens": 59,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34835,
            "output_tokens": 59,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35122,
            "output_tokens": 537,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35682,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 39,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 14,
          "bash": 1,
          "exec": 15,
          "devices": 3,
          "device": 1,
          "write": 2,
          "device_stop": 1,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 246 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/phone_stand.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/workspace/.kinex/memos/phone_stand.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-41-20261113/workspace/.kinex/tools/tcp_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 76,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 100,
        "published_frames": 100,
        "decode_ok": true,
        "duration_s": 2.5,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "4a0b03d248866ac0d6a5cf5bf0d9c3e977dce9bc02402ba194917ca4c0250fc5",
        "source_sha256": {
          "third_person": "708753e31b7df54ca3e0d8bb0efb9ff1f6587d5837ec27e170b709605cd10516",
          "left_wrist": "9fbbe64638314f8c52b9f7f0674c1f137f5c202cd422801b1f313c63cc463c68",
          "right_wrist": "8842e2f1506f191c0fb255eea2fe0eba692fec21f1a5070737a1890f20ac56b4"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "c176b14474e6",
          "task": "robotwin/place-phone-stand-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 100,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 100,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9840000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 100,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 100,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9840000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 100,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 100,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9840000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-42-20261113",
      "variant": "kinex-native",
      "task_key": "task03/42",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "42",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 8.2,
        "step_limit": 500,
        "steps": 205,
        "success": true,
        "valid": true
      },
      "steps": 205,
      "simulation_time_s": 8.2,
      "wall_time_s": 731.804742,
      "timing": {
        "simulation_time_s": 8.2,
        "wall_time_s": 731.804742,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the casual brown shoe off the table and drop it on the mat",
      "native_instruction": "Grab the casual brown shoe off the table and drop it on the mat",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9661779179810726,
        "cache_read_input_tokens": 765696,
        "cache_reported_input_tokens": 792500,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 792500,
        "cached_input_tokens": 765696,
        "cost_usd": null,
        "fresh_input_tokens": 26804,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 792500,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 765696,
        "known_fresh_input_tokens": 26804,
        "known_input_tokens": 792500,
        "known_output_tokens": 4364,
        "known_reasoning_output_tokens": 626,
        "output_tokens": 4364,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 626,
        "reasoning_reported_output_tokens": 4364,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 34,
          "cache_write_input_tokens": 34,
          "cache_write_reported_input_tokens": 34,
          "cached_input_tokens": 34,
          "fresh_input_tokens": 34,
          "input_tokens": 34,
          "output_tokens": 34,
          "reasoning_output_tokens": 34,
          "reasoning_reported_output_tokens": 34
        },
        "request_attempts": 34,
        "response_count": 34,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 796864,
        "uncached_input_tokens": 26804,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11850,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11981,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12280,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12743,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12800,
            "input_tokens": 16137,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 17400,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 17564,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 17755,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 18325,
            "output_tokens": 208,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 20689,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 20791,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21569,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 21977,
            "output_tokens": 275,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22787,
            "output_tokens": 258,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23759,
            "output_tokens": 185,
            "reasoning_output_tokens": 143
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 25123,
            "output_tokens": 654,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25798,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25959,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26233,
            "output_tokens": 154,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26447,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26536,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26952,
            "output_tokens": 112,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27094,
            "output_tokens": 163,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27284,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27951,
            "output_tokens": 148,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28784,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29308,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29995,
            "output_tokens": 278,
            "reasoning_output_tokens": 186
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30896,
            "output_tokens": 180,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31763,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31960,
            "output_tokens": 61,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32249,
            "output_tokens": 473,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32746,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 40,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 14,
          "bash": 1,
          "exec": 15,
          "devices": 4,
          "device": 1,
          "write": 2,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 205 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-shoe.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/workspace/.kinex/memos/robotwin-shoe.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-42-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 77,
        "observed_images": 9
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 83,
        "published_frames": 83,
        "decode_ok": true,
        "duration_s": 2.075,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "dc4275a506eb2c989bfee4683cf28aa485574e214ca2a142c67c3b8b7275b520",
        "source_sha256": {
          "third_person": "e9b27bf3380ceab4c403e078278a72304e822f212373c5784264693c49db5738",
          "left_wrist": "e3c75b0f3274a35b71b662d5996fa479b1937ec57596332559151924bb3d6714",
          "right_wrist": "dec64b8ab0a2cdc9408b8a2fd71686d42f5a93803fc866d2f71540b51504a746"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "bf36ef6132e5",
          "task": "robotwin/place-shoe-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 84,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 83,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 84,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 83,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 84,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 83,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-43-20261113",
      "variant": "kinex-native",
      "task_key": "task03/43",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "43",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 3.7600000000000002,
        "step_limit": 400,
        "steps": 94,
        "success": true,
        "valid": true
      },
      "steps": 94,
      "simulation_time_s": 3.7600000000000002,
      "wall_time_s": 398.896927,
      "timing": {
        "simulation_time_s": 3.7600000000000002,
        "wall_time_s": 398.896927,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to press the shiny red plastic stapler.",
      "native_instruction": "Use the left arm to press the shiny red plastic stapler.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9466804923455743,
        "cache_read_input_tokens": 633600,
        "cache_reported_input_tokens": 669286,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 669286,
        "cached_input_tokens": 633600,
        "cost_usd": null,
        "fresh_input_tokens": 35686,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 669286,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 633600,
        "known_fresh_input_tokens": 35686,
        "known_input_tokens": 669286,
        "known_output_tokens": 4048,
        "known_reasoning_output_tokens": 391,
        "output_tokens": 4048,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 391,
        "reasoning_reported_output_tokens": 4048,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 27,
          "cache_write_input_tokens": 27,
          "cache_write_reported_input_tokens": 27,
          "cached_input_tokens": 27,
          "fresh_input_tokens": 27,
          "input_tokens": 27,
          "output_tokens": 27,
          "reasoning_output_tokens": 27,
          "reasoning_reported_output_tokens": 27
        },
        "request_attempts": 27,
        "response_count": 27,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 673334,
        "uncached_input_tokens": 35686,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11991,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 16232,
            "output_tokens": 43,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16128,
            "input_tokens": 16328,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 19366,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 20628,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20792,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 22222,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22474,
            "output_tokens": 340,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23209,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23617,
            "output_tokens": 311,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24463,
            "output_tokens": 352,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25741,
            "output_tokens": 157,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 28234,
            "output_tokens": 946,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29202,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29705,
            "output_tokens": 65,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29830,
            "output_tokens": 50,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29902,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30391,
            "output_tokens": 95,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30500,
            "output_tokens": 145,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30673,
            "output_tokens": 65,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31182,
            "output_tokens": 93,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31723,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31943,
            "output_tokens": 75,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32358,
            "output_tokens": 522,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 32904,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 34,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 12,
          "find": 1,
          "bash": 1,
          "exec": 10,
          "devices": 3,
          "device": 1,
          "write": 2,
          "device_stop": 2,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 94 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/press-stapler.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/workspace/.kinex/memos/press-stapler.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/stapler_motion.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-43-20261113/workspace/.kinex/tools/stapler_motion.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 64,
        "observed_images": 4
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 39,
        "published_frames": 39,
        "decode_ok": true,
        "duration_s": 0.975,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "0157fed526b13fd012158f2278f67055fb7046e61f106b03299992b187a8fb61",
        "source_sha256": {
          "third_person": "6042e4afb4eaa79a6acfa6ba4077e8abf3ba9662451c3477f7f4036acbd07008",
          "left_wrist": "4b9474f3bc19bc10c42fba5765f559f672cfc039b50ec6d9eed30a25f114777e",
          "right_wrist": "dc5d86cd2bea1fe317afd7a7560d066a2876eff48f5deabebe230eefcaa37dc0"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "36c92e46e9ca",
          "task": "robotwin/press-stapler-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 39,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 39,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3760000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 39,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 39,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3760000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 39,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 39,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3760000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-44-20261113",
      "variant": "kinex-native",
      "task_key": "task03/44",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "44",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.6,
        "step_limit": 400,
        "steps": 315,
        "success": true,
        "valid": true
      },
      "steps": 315,
      "simulation_time_s": 12.6,
      "wall_time_s": 1284.759853,
      "timing": {
        "simulation_time_s": 12.6,
        "wall_time_s": 1284.759853,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Catch the tabletop paymentsign, pick it up and rotate it, then set it back down.",
      "native_instruction": "Catch the tabletop paymentsign, pick it up and rotate it, then set it back down.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9634829163743722,
        "cache_read_input_tokens": 1567104,
        "cache_reported_input_tokens": 1626499,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1626499,
        "cached_input_tokens": 1567104,
        "cost_usd": null,
        "fresh_input_tokens": 59395,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1626499,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1567104,
        "known_fresh_input_tokens": 59395,
        "known_input_tokens": 1626499,
        "known_output_tokens": 9342,
        "known_reasoning_output_tokens": 2543,
        "output_tokens": 9342,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2543,
        "reasoning_reported_output_tokens": 9342,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 51,
          "cache_write_input_tokens": 51,
          "cache_write_reported_input_tokens": 51,
          "cached_input_tokens": 51,
          "fresh_input_tokens": 51,
          "input_tokens": 51,
          "output_tokens": 51,
          "reasoning_output_tokens": 51,
          "reasoning_reported_output_tokens": 51
        },
        "request_attempts": 51,
        "response_count": 51,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1635841,
        "uncached_input_tokens": 59395,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11822,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11862,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11995,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 14836,
            "output_tokens": 211,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 18715,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 19990,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20220,
            "output_tokens": 138,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 22285,
            "output_tokens": 78,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22573,
            "output_tokens": 134,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 23645,
            "output_tokens": 310,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24802,
            "output_tokens": 340,
            "reasoning_output_tokens": 298
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 26321,
            "output_tokens": 358,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26935,
            "output_tokens": 389,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27654,
            "output_tokens": 464,
            "reasoning_output_tokens": 344
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28387,
            "output_tokens": 1185,
            "reasoning_output_tokens": 288
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29594,
            "output_tokens": 137,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29862,
            "output_tokens": 137,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30643,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30763,
            "output_tokens": 137,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30960,
            "output_tokens": 58,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31040,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31295,
            "output_tokens": 145,
            "reasoning_output_tokens": 110
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31546,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31707,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32574,
            "output_tokens": 57,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32653,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32748,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33130,
            "output_tokens": 133,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33692,
            "output_tokens": 69,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 34132,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34633,
            "output_tokens": 198,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34912,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35716,
            "output_tokens": 57,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36544,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36674,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37234,
            "output_tokens": 106,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37362,
            "output_tokens": 291,
            "reasoning_output_tokens": 191
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38401,
            "output_tokens": 83,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 39179,
            "output_tokens": 148,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39355,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39638,
            "output_tokens": 209,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40276,
            "output_tokens": 74,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40372,
            "output_tokens": 545,
            "reasoning_output_tokens": 227
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41326,
            "output_tokens": 412,
            "reasoning_output_tokens": 143
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42511,
            "output_tokens": 265,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 42804,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43330,
            "output_tokens": 216,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43949,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 44141,
            "output_tokens": 152,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44543,
            "output_tokens": 650,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45218,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 59,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 14,
          "find": 1,
          "exec": 26,
          "devices": 2,
          "device": 1,
          "write": 2,
          "edit": 6,
          "device_stop": 4,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 315 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin_manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/workspace/.kinex/memos/robotwin_manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/robotwin_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-44-20261113/workspace/.kinex/tools/robotwin_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 113,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 127,
        "published_frames": 127,
        "decode_ok": true,
        "duration_s": 3.175,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "6a1d65af2e2c181ccae98fd9b22606058bc9fbeab2d8c2461087394b138f66d7",
        "source_sha256": {
          "third_person": "757ed2e262ad40c7b0aa3809a14abce40d64e17bddfb0b95e86f20855a9deaee",
          "left_wrist": "840f0ab5166598144c9db2d3a222fd4675673459ada4916947a7e53f7b855af2",
          "right_wrist": "3b03340c84f1769dfd59027619b5a881bd7a34af474db9d0cc02c2a65ac8fe43"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "f02a39f0f288",
          "task": "robotwin/rotate-qrcode-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 128,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 127,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 128,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 127,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 128,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 127,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-45-20261113",
      "variant": "kinex-native",
      "task_key": "task03/45",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "45",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 5.84,
        "step_limit": 700,
        "steps": 146,
        "success": true,
        "valid": true
      },
      "steps": 146,
      "simulation_time_s": 5.84,
      "wall_time_s": 677.985858,
      "timing": {
        "simulation_time_s": 5.84,
        "wall_time_s": 677.985858,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the bottle with white cap with the left arm and shake.",
      "native_instruction": "Pick up the bottle with white cap with the left arm and shake.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9706258951182267,
        "cache_read_input_tokens": 614016,
        "cache_reported_input_tokens": 632598,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 632598,
        "cached_input_tokens": 614016,
        "cost_usd": null,
        "fresh_input_tokens": 18582,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 632598,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 614016,
        "known_fresh_input_tokens": 18582,
        "known_input_tokens": 632598,
        "known_output_tokens": 3916,
        "known_reasoning_output_tokens": 485,
        "output_tokens": 3916,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 485,
        "reasoning_reported_output_tokens": 3916,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 30,
          "cache_write_input_tokens": 30,
          "cache_write_reported_input_tokens": 30,
          "cached_input_tokens": 30,
          "fresh_input_tokens": 30,
          "input_tokens": 30,
          "output_tokens": 30,
          "reasoning_output_tokens": 30,
          "reasoning_reported_output_tokens": 30
        },
        "request_attempts": 30,
        "response_count": 30,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 636514,
        "uncached_input_tokens": 18582,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11862,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11995,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12294,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 15332,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15232,
            "input_tokens": 16595,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 16759,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 17329,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 17784,
            "output_tokens": 99,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 18093,
            "output_tokens": 134,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 19015,
            "output_tokens": 292,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 19915,
            "output_tokens": 221,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 21779,
            "output_tokens": 732,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22536,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22699,
            "output_tokens": 111,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 22870,
            "output_tokens": 53,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 22945,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23201,
            "output_tokens": 92,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23504,
            "output_tokens": 160,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 23693,
            "output_tokens": 105,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24301,
            "output_tokens": 201,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 25273,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25330,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25711,
            "output_tokens": 60,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25877,
            "output_tokens": 152,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26058,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26433,
            "output_tokens": 105,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26680,
            "output_tokens": 88,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27165,
            "output_tokens": 564,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 27752,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 34,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 10,
          "bash": 1,
          "exec": 9,
          "devices": 6,
          "write": 2,
          "device_stop": 2,
          "edit": 2,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 146 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/shake-bottle.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/workspace/.kinex/memos/shake-bottle.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/left_bottle_motion.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-45-20261113/workspace/.kinex/tools/left_bottle_motion.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 67,
        "observed_images": 5
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 60,
        "published_frames": 60,
        "decode_ok": true,
        "duration_s": 1.5,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "80e1bb98bf359f435e774e6309cb596f4d09aea2c4e7828915261e5bb00c4471",
        "source_sha256": {
          "third_person": "726afd6a93fb2816bdf76af0a16312c33e143de7119cebfae37c5bba873b12bb",
          "left_wrist": "0d0a2355b719d4cbde867e0bd897950b1af62c56077a9c2a7c8134e40530a569",
          "right_wrist": "83f02b055e03f1118239723dffc20ad7520ced8c0f80037a226f92a0bf3f6484"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "879549c35f03",
          "task": "robotwin/shake-bottle-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 60,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 60,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 5840000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 60,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 60,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 5840000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 60,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 60,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 5840000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-46-20261113",
      "variant": "kinex-native",
      "task_key": "task03/46",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "46",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 6.32,
        "step_limit": 700,
        "steps": 158,
        "success": true,
        "valid": true
      },
      "steps": 158,
      "simulation_time_s": 6.32,
      "wall_time_s": 475.730331,
      "timing": {
        "simulation_time_s": 6.32,
        "wall_time_s": 475.730331,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the bottle with white cap and shake it horizontally.",
      "native_instruction": "Grab the bottle with white cap and shake it horizontally.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9465495762070917,
        "cache_read_input_tokens": 556928,
        "cache_reported_input_tokens": 588377,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 588377,
        "cached_input_tokens": 556928,
        "cost_usd": null,
        "fresh_input_tokens": 31449,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 588377,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 556928,
        "known_fresh_input_tokens": 31449,
        "known_input_tokens": 588377,
        "known_output_tokens": 4082,
        "known_reasoning_output_tokens": 495,
        "output_tokens": 4082,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 495,
        "reasoning_reported_output_tokens": 4082,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 26,
          "cache_write_input_tokens": 26,
          "cache_write_reported_input_tokens": 26,
          "cached_input_tokens": 26,
          "fresh_input_tokens": 26,
          "input_tokens": 26,
          "output_tokens": 26,
          "reasoning_output_tokens": 26,
          "reasoning_reported_output_tokens": 26
        },
        "request_attempts": 26,
        "response_count": 26,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 592459,
        "uncached_input_tokens": 31449,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11816,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11856,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11989,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12039,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12287,
            "output_tokens": 204,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 16150,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 17426,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 17656,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16128,
            "input_tokens": 19085,
            "output_tokens": 151,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 20575,
            "output_tokens": 333,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 22174,
            "output_tokens": 173,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 24683,
            "output_tokens": 1117,
            "reasoning_output_tokens": 144
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25823,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25977,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26212,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26363,
            "output_tokens": 49,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26474,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26884,
            "output_tokens": 177,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27090,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27829,
            "output_tokens": 170,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28637,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29268,
            "output_tokens": 127,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29537,
            "output_tokens": 61,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29660,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30149,
            "output_tokens": 563,
            "reasoning_output_tokens": 92
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30738,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 36,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 14,
          "bash": 1,
          "exec": 10,
          "devices": 3,
          "device": 1,
          "write": 2,
          "device_stop": 2,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 158 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/aloha_bottle.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/workspace/.kinex/memos/aloha_bottle.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-46-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 65,
        "observed_images": 6
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 64,
        "published_frames": 64,
        "decode_ok": true,
        "duration_s": 1.6,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "a40c9b2b69ccb4f0b65d57c44797bce734b88384c154476a3771bc1788acae71",
        "source_sha256": {
          "third_person": "28388b62a81e3bc192dabea963bfb7aa0afe48804476296acfb67e9d438607e7",
          "left_wrist": "faf1a14fdb7594452b720f70932ef2917ef5004c4d7735344d81896d53a1615a",
          "right_wrist": "fe37d66646c1d4de22703ce06cfb52ba33c1cf998206237bd479440e5a368755"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "f74937815c2e",
          "task": "robotwin/shake-bottle-horizontally-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 65,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 64,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 6320000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 65,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 64,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 6320000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 65,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 64,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 6320000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-47-20261113",
      "variant": "kinex-native",
      "task_key": "task03/47",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "47",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 15.44,
        "step_limit": 800,
        "steps": 386,
        "success": true,
        "valid": true
      },
      "steps": 386,
      "simulation_time_s": 15.44,
      "wall_time_s": 1136.158109,
      "timing": {
        "simulation_time_s": 15.44,
        "wall_time_s": 1136.158109,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Bring red block and green block to the center, then position green block above red block.",
      "native_instruction": "Bring red block and green block to the center, then position green block above red block.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9663287631638927,
        "cache_read_input_tokens": 1344000,
        "cache_reported_input_tokens": 1390831,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1390831,
        "cached_input_tokens": 1344000,
        "cost_usd": null,
        "fresh_input_tokens": 46831,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1390831,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1344000,
        "known_fresh_input_tokens": 46831,
        "known_input_tokens": 1390831,
        "known_output_tokens": 6585,
        "known_reasoning_output_tokens": 1062,
        "output_tokens": 6585,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1062,
        "reasoning_reported_output_tokens": 6585,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 49,
          "cache_write_input_tokens": 49,
          "cache_write_reported_input_tokens": 49,
          "cached_input_tokens": 49,
          "fresh_input_tokens": 49,
          "input_tokens": 49,
          "output_tokens": 49,
          "reasoning_output_tokens": 49,
          "reasoning_reported_output_tokens": 49
        },
        "request_attempts": 49,
        "response_count": 49,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1397416,
        "uncached_input_tokens": 46831,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11867,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11998,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13828,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13696,
            "input_tokens": 13911,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13824,
            "input_tokens": 14062,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14080,
            "input_tokens": 18588,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 19158,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 20420,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20650,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20842,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21094,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21502,
            "output_tokens": 479,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 22663,
            "output_tokens": 44,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22769,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23434,
            "output_tokens": 166,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23746,
            "output_tokens": 67,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24992,
            "output_tokens": 1220,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26233,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26987,
            "output_tokens": 86,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27100,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27774,
            "output_tokens": 288,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28089,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28858,
            "output_tokens": 72,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29625,
            "output_tokens": 223,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30190,
            "output_tokens": 93,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30383,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30526,
            "output_tokens": 49,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30637,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 31302,
            "output_tokens": 111,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31450,
            "output_tokens": 235,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31712,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32615,
            "output_tokens": 78,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32798,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33707,
            "output_tokens": 124,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34512,
            "output_tokens": 150,
            "reasoning_output_tokens": 106
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35633,
            "output_tokens": 119,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35779,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36565,
            "output_tokens": 263,
            "reasoning_output_tokens": 172
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37705,
            "output_tokens": 73,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37887,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38857,
            "output_tokens": 94,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39209,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39371,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 40277,
            "output_tokens": 177,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 41328,
            "output_tokens": 89,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41559,
            "output_tokens": 79,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 42038,
            "output_tokens": 721,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 42782,
            "output_tokens": 76,
            "reasoning_output_tokens": 42
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 53,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 14,
          "find": 1,
          "bash": 1,
          "exec": 24,
          "devices": 2,
          "device": 2,
          "device_stop": 2,
          "write": 2,
          "edit": 4
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 386 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack-blocks.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/workspace/.kinex/memos/stack-blocks.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/stack_blocks.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-47-20261113/workspace/.kinex/tools/stack_blocks.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 105,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 156,
        "published_frames": 156,
        "decode_ok": true,
        "duration_s": 3.9,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f299ccac7c4c0f8ca4d1b9c02cc3f7dd4ecc4777e4b2a45e1050f19c13e76bd3",
        "source_sha256": {
          "third_person": "5d6f093b98f1e12f95e7276b73f149f690e41bc1cf452a09af40faa4b4671703",
          "left_wrist": "b3f16db2935ea532dca7455aded514c3862b3ee34ca5167c29862c04c1c46a62",
          "right_wrist": "a2c1b5918686245ef5c172ca47142650770dd8db2c4f7fa22928c15e29a4f3d2"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "7182fc1b70e0",
          "task": "robotwin/stack-blocks-two-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 156,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 156,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15440000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 156,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 156,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15440000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 156,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 156,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15440000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-48-20261113",
      "variant": "kinex-native",
      "task_key": "task03/48",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "48",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 29.44,
        "step_limit": 900,
        "steps": 736,
        "success": true,
        "valid": true
      },
      "steps": 736,
      "simulation_time_s": 29.44,
      "wall_time_s": 1476.929241,
      "timing": {
        "simulation_time_s": 29.44,
        "wall_time_s": 1476.929241,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Stack the small smooth brown-rimmed bowl directly over the smooth bowl with glossy finish.",
      "native_instruction": "Stack the small smooth brown-rimmed bowl directly over the smooth bowl with glossy finish.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9816975317758021,
        "cache_read_input_tokens": 2093952,
        "cache_reported_input_tokens": 2132991,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2132991,
        "cached_input_tokens": 2093952,
        "cost_usd": null,
        "fresh_input_tokens": 39039,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2132991,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2093952,
        "known_fresh_input_tokens": 39039,
        "known_input_tokens": 2132991,
        "known_output_tokens": 8564,
        "known_reasoning_output_tokens": 2098,
        "output_tokens": 8564,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2098,
        "reasoning_reported_output_tokens": 8564,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 60,
          "cache_write_input_tokens": 60,
          "cache_write_reported_input_tokens": 60,
          "cached_input_tokens": 60,
          "fresh_input_tokens": 60,
          "input_tokens": 60,
          "output_tokens": 60,
          "reasoning_output_tokens": 60,
          "reasoning_reported_output_tokens": 60
        },
        "request_attempts": 60,
        "response_count": 60,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2141555,
        "uncached_input_tokens": 39039,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11824,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11864,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11997,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 16127,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 16386,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 20235,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20096,
            "input_tokens": 21497,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 21727,
            "output_tokens": 104,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 23292,
            "output_tokens": 196,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 25603,
            "output_tokens": 376,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26661,
            "output_tokens": 193,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 28025,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28216,
            "output_tokens": 230,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 29036,
            "output_tokens": 188,
            "reasoning_output_tokens": 148
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29690,
            "output_tokens": 188,
            "reasoning_output_tokens": 149
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29941,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30302,
            "output_tokens": 334,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30859,
            "output_tokens": 178,
            "reasoning_output_tokens": 136
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 32216,
            "output_tokens": 855,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33092,
            "output_tokens": 217,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33562,
            "output_tokens": 188,
            "reasoning_output_tokens": 149
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34445,
            "output_tokens": 336,
            "reasoning_output_tokens": 121
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35038,
            "output_tokens": 398,
            "reasoning_output_tokens": 284
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35748,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36145,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36338,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36696,
            "output_tokens": 143,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37156,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37349,
            "output_tokens": 207,
            "reasoning_output_tokens": 112
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37873,
            "output_tokens": 58,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 37953,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38007,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38200,
            "output_tokens": 86,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38591,
            "output_tokens": 191,
            "reasoning_output_tokens": 112
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39100,
            "output_tokens": 111,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39528,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39933,
            "output_tokens": 63,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40018,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40380,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40775,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 40966,
            "output_tokens": 156,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41440,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41844,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42205,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 43047,
            "output_tokens": 104,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 43466,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43658,
            "output_tokens": 431,
            "reasoning_output_tokens": 190
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44523,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44715,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45117,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45483,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45880,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 46071,
            "output_tokens": 289,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46450,
            "output_tokens": 263,
            "reasoning_output_tokens": 174
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 47031,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47223,
            "output_tokens": 133,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47678,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 47877,
            "output_tokens": 40,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47872,
            "input_tokens": 48087,
            "output_tokens": 694,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48640,
            "input_tokens": 48805,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 69,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 16,
          "find": 1,
          "bash": 1,
          "exec": 31,
          "devices": 3,
          "device": 11,
          "write": 2,
          "device_stop": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 736 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack_bowls.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/workspace/.kinex/memos/stack_bowls.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/bowl_motion.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-48-20261113/workspace/.kinex/tools/bowl_motion.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 132,
        "observed_images": 13
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 296,
        "published_frames": 296,
        "decode_ok": true,
        "duration_s": 7.4,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "7abfde869cb7e656f7a3ddde34a7d7be793ff7d2f9e9c961a13470c3c6333867",
        "source_sha256": {
          "third_person": "030617822dc4a5d3b7e2fb53f7391ebe7f4906061a7abe84893ddbe79f77b37a",
          "left_wrist": "bcb5d714f3107d62c29b437a4f0eecf975f10e6b0dbb87ab6cf9bcf9b00d98cd",
          "right_wrist": "a3a3ab158c445e5dbc2b1dd7f1a3007a3f84d2030ef506f7e34b86545aaaffeb"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "842d9bbfd502",
          "task": "robotwin/stack-bowls-two-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 296,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 296,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 29440000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 296,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 296,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 29440000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 296,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 296,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 29440000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-49-20261113",
      "variant": "kinex-native",
      "task_key": "task03/49",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "49",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 11.16,
        "step_limit": 400,
        "steps": 279,
        "success": true,
        "valid": true
      },
      "steps": 279,
      "simulation_time_s": 11.16,
      "wall_time_s": 1138.93272,
      "timing": {
        "simulation_time_s": 11.16,
        "wall_time_s": 1138.93272,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Stamp Green using the block-shaped seal with angled edges",
      "native_instruction": "Stamp Green using the block-shaped seal with angled edges",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9389324120480755,
        "cache_read_input_tokens": 926208,
        "cache_reported_input_tokens": 986448,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 986448,
        "cached_input_tokens": 926208,
        "cost_usd": null,
        "fresh_input_tokens": 60240,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 986448,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 926208,
        "known_fresh_input_tokens": 60240,
        "known_input_tokens": 986448,
        "known_output_tokens": 5829,
        "known_reasoning_output_tokens": 1101,
        "output_tokens": 5829,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1101,
        "reasoning_reported_output_tokens": 5829,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 38,
          "cache_write_input_tokens": 38,
          "cache_write_reported_input_tokens": 38,
          "cached_input_tokens": 38,
          "fresh_input_tokens": 38,
          "input_tokens": 38,
          "output_tokens": 38,
          "reasoning_output_tokens": 38,
          "reasoning_reported_output_tokens": 38
        },
        "request_attempts": 38,
        "response_count": 38,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 992277,
        "uncached_input_tokens": 60240,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11988,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12036,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12284,
            "output_tokens": 192,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 16135,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 16705,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 17980,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18176,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 18369,
            "output_tokens": 168,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 19578,
            "output_tokens": 328,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 21246,
            "output_tokens": 362,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 22098,
            "output_tokens": 780,
            "reasoning_output_tokens": 127
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 22899,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23016,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23732,
            "output_tokens": 73,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 23832,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 23985,
            "output_tokens": 41,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24075,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 25294,
            "output_tokens": 104,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 28390,
            "output_tokens": 74,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28637,
            "output_tokens": 103,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 29318,
            "output_tokens": 120,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 30048,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30717,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31371,
            "output_tokens": 506,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 32439,
            "output_tokens": 252,
            "reasoning_output_tokens": 160
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33258,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33921,
            "output_tokens": 125,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34621,
            "output_tokens": 165,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35370,
            "output_tokens": 98,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 36041,
            "output_tokens": 321,
            "reasoning_output_tokens": 228
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36938,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37174,
            "output_tokens": 59,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37282,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37474,
            "output_tokens": 387,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37888,
            "output_tokens": 552,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38463,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 45,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 13,
          "bash": 1,
          "exec": 18,
          "devices": 3,
          "device": 2,
          "write": 2,
          "edit": 2,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 279 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stamp-seal.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/workspace/.kinex/memos/stamp-seal.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-49-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 86,
        "observed_images": 15
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 113,
        "published_frames": 113,
        "decode_ok": true,
        "duration_s": 2.825,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "12f5b3ecf977aba365a73dbd8e653bc1fc504e102b241eda57140188a50ba730",
        "source_sha256": {
          "third_person": "39b359898022b5c9b1d5349d838abc91c4048fe953710c1be8fbcab6c3496274",
          "left_wrist": "5a5e26279e1e2b8de3c259ce272dda421e4cb35fc788883b8c8b3137a5972630",
          "right_wrist": "bc2849df00869873cf26ecab185c413d58aa1a946d82069381fbddabf40096bb"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "75dd8a6922fc",
          "task": "robotwin/stamp-seal-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 113,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 113,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11160000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 113,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 113,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11160000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 113,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 113,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11160000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-50-20261113",
      "variant": "kinex-native",
      "task_key": "task03/50",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "50",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 14.72,
        "step_limit": 400,
        "steps": 368,
        "success": false,
        "valid": true
      },
      "steps": 368,
      "simulation_time_s": 14.72,
      "wall_time_s": 1344.85536,
      "timing": {
        "simulation_time_s": 14.72,
        "wall_time_s": 1344.85536,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Click the flat gray switch with the right arm",
      "native_instruction": "Click the flat gray switch with the right arm",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9641827775709938,
        "cache_read_input_tokens": 1100928,
        "cache_reported_input_tokens": 1141825,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1141825,
        "cached_input_tokens": 1100928,
        "cost_usd": null,
        "fresh_input_tokens": 40897,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1141825,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1100928,
        "known_fresh_input_tokens": 40897,
        "known_input_tokens": 1141825,
        "known_output_tokens": 7496,
        "known_reasoning_output_tokens": 2595,
        "output_tokens": 7496,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2595,
        "reasoning_reported_output_tokens": 7496,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 42,
          "cache_write_input_tokens": 42,
          "cache_write_reported_input_tokens": 42,
          "cached_input_tokens": 42,
          "fresh_input_tokens": 42,
          "input_tokens": 42,
          "output_tokens": 42,
          "reasoning_output_tokens": 42,
          "reasoning_reported_output_tokens": 42
        },
        "request_attempts": 42,
        "response_count": 42,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1149321,
        "uncached_input_tokens": 40897,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11813,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11853,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11986,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12034,
            "output_tokens": 134,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 12376,
            "output_tokens": 165,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16072,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 17334,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 17904,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18100,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18291,
            "output_tokens": 211,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 20149,
            "output_tokens": 330,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 21988,
            "output_tokens": 150,
            "reasoning_output_tokens": 112
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22833,
            "output_tokens": 538,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 24145,
            "output_tokens": 357,
            "reasoning_output_tokens": 315
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 25681,
            "output_tokens": 888,
            "reasoning_output_tokens": 209
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26589,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26966,
            "output_tokens": 106,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27132,
            "output_tokens": 39,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27209,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27623,
            "output_tokens": 305,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 27954,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28667,
            "output_tokens": 111,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29398,
            "output_tokens": 76,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29703,
            "output_tokens": 415,
            "reasoning_output_tokens": 335
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30883,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 30950,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31530,
            "output_tokens": 276,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31832,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31973,
            "output_tokens": 136,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32793,
            "output_tokens": 212,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33270,
            "output_tokens": 135,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 34005,
            "output_tokens": 51,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34227,
            "output_tokens": 140,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34870,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35063,
            "output_tokens": 303,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 36021,
            "output_tokens": 895,
            "reasoning_output_tokens": 775
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37521,
            "output_tokens": 44,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37735,
            "output_tokens": 187,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38534,
            "output_tokens": 55,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38611,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38804,
            "output_tokens": 577,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39403,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 52,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 13,
          "find": 4,
          "exec": 16,
          "devices": 4,
          "device": 5,
          "write": 2,
          "device_stop": 3,
          "edit": 2,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 368 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/turn-switch.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/workspace/.kinex/memos/turn-switch.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/right_tcp.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-50-20261113/workspace/.kinex/tools/right_tcp.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 97,
        "observed_images": 13
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113"
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 148,
        "published_frames": 148,
        "decode_ok": true,
        "duration_s": 3.7,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "beb7611e634b80c48bf60a5910de1aff03002066d64046fb16bfbd2cb62cd783",
        "source_sha256": {
          "third_person": "de16064781f714a7d43b7d48a40cd90ce0bf3366d4c66d9fe827fba699ddf737",
          "left_wrist": "5d9eeb042dbf87a8149b117f6443f96c68cf8b36a94421c99dcfd373a9ec0384",
          "right_wrist": "881c58cab51d0235541a2e155ce9e95daf2a4e137df8bcbe219c574cc24c755a"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "8016e9dcad43",
          "task": "robotwin/turn-switch-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 149,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 148,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14720000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 149,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 148,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14720000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 149,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 148,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14720000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-01-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/01",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "01",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 28.28,
        "step_limit": 800,
        "steps": 707,
        "success": false,
        "valid": true
      },
      "steps": 707,
      "simulation_time_s": 28.28,
      "wall_time_s": 3322.226984,
      "timing": {
        "simulation_time_s": 28.28,
        "wall_time_s": 3322.226984,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "native_instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9669555545256523,
        "cache_read_input_tokens": 3087872,
        "cache_reported_input_tokens": 3193396,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 3193396,
        "cached_input_tokens": 3087872,
        "cost_usd": null,
        "fresh_input_tokens": 105524,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 3193396,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 3087872,
        "known_fresh_input_tokens": 105524,
        "known_input_tokens": 3193396,
        "known_output_tokens": 13158,
        "known_reasoning_output_tokens": 3753,
        "output_tokens": 13158,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3753,
        "reasoning_reported_output_tokens": 13158,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 82,
          "cache_write_input_tokens": 82,
          "cache_write_reported_input_tokens": 82,
          "cached_input_tokens": 82,
          "fresh_input_tokens": 82,
          "input_tokens": 82,
          "output_tokens": 82,
          "reasoning_output_tokens": 82,
          "reasoning_reported_output_tokens": 82
        },
        "request_attempts": 82,
        "response_count": 82,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 3206554,
        "uncached_input_tokens": 105524,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11828,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11868,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12001,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12051,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13631,
            "output_tokens": 236,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13696,
            "input_tokens": 18059,
            "output_tokens": 169,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 21101,
            "output_tokens": 116,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 22331,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22657,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22847,
            "output_tokens": 66,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23282,
            "output_tokens": 629,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 23934,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24125,
            "output_tokens": 44,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24231,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24354,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25182,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 26016,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26107,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26665,
            "output_tokens": 150,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26844,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26949,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27042,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27283,
            "output_tokens": 102,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 28171,
            "output_tokens": 65,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28446,
            "output_tokens": 415,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29364,
            "output_tokens": 367,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30074,
            "output_tokens": 120,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30660,
            "output_tokens": 853,
            "reasoning_output_tokens": 185
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31536,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32172,
            "output_tokens": 124,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32356,
            "output_tokens": 40,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32458,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33054,
            "output_tokens": 145,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 34095,
            "output_tokens": 126,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 35004,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35101,
            "output_tokens": 110,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35240,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35480,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36120,
            "output_tokens": 72,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36624,
            "output_tokens": 107,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37282,
            "output_tokens": 305,
            "reasoning_output_tokens": 191
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 38138,
            "output_tokens": 143,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38969,
            "output_tokens": 71,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39852,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 39968,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40774,
            "output_tokens": 206,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41871,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42498,
            "output_tokens": 68,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42777,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43139,
            "output_tokens": 379,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43911,
            "output_tokens": 383,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44652,
            "output_tokens": 330,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45184,
            "output_tokens": 462,
            "reasoning_output_tokens": 379
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 46192,
            "output_tokens": 86,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46489,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46980,
            "output_tokens": 116,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47532,
            "output_tokens": 118,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47360,
            "input_tokens": 48206,
            "output_tokens": 167,
            "reasoning_output_tokens": 86
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 48922,
            "output_tokens": 122,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49106,
            "output_tokens": 314,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49280,
            "input_tokens": 50210,
            "output_tokens": 184,
            "reasoning_output_tokens": 143
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50304,
            "input_tokens": 51089,
            "output_tokens": 464,
            "reasoning_output_tokens": 334
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51456,
            "input_tokens": 52333,
            "output_tokens": 176,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52352,
            "input_tokens": 53064,
            "output_tokens": 71,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52992,
            "input_tokens": 53565,
            "output_tokens": 158,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53504,
            "input_tokens": 54277,
            "output_tokens": 412,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54528,
            "input_tokens": 55619,
            "output_tokens": 222,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55680,
            "input_tokens": 56392,
            "output_tokens": 143,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56320,
            "input_tokens": 56557,
            "output_tokens": 84,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56448,
            "input_tokens": 57079,
            "output_tokens": 114,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57088,
            "input_tokens": 57738,
            "output_tokens": 362,
            "reasoning_output_tokens": 320
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57984,
            "input_tokens": 58122,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58112,
            "input_tokens": 58263,
            "output_tokens": 179,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58368,
            "input_tokens": 58997,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59008,
            "input_tokens": 59641,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59520,
            "input_tokens": 60132,
            "output_tokens": 127,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60160,
            "input_tokens": 60810,
            "output_tokens": 77,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60672,
            "input_tokens": 61211,
            "output_tokens": 349,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61312,
            "input_tokens": 61839,
            "output_tokens": 254,
            "reasoning_output_tokens": 208
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 62155,
            "output_tokens": 238,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62208,
            "input_tokens": 62422,
            "output_tokens": 649,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62848,
            "input_tokens": 63096,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 91,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 17,
          "find": 1,
          "exec": 45,
          "devices": 7,
          "device": 1,
          "write": 3,
          "device_stop": 9,
          "edit": 5
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 707 rosbridge control frames; 49 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/handover-block.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/workspace/.kinex/memos/handover-block.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/workspace/.kinex/tools/cartesian_step.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_batch.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-01-20261113/workspace/.kinex/tools/policy_batch.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 176,
        "observed_images": 43
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 49,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 284,
        "published_frames": 284,
        "decode_ok": true,
        "duration_s": 7.1,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "1e3cec9083d5b79cf9f31fe0547c1fee4d45f0c7ce3ed304802d5c84051d9bcd",
        "source_sha256": {
          "third_person": "e6e1702b74727749fabaf4a78bfca1d2ac0e38cdfce39ef2dfae4519a0c50499",
          "left_wrist": "0a098c1258f5273c81bf894d3bba91ef03e1991e5f97a08670f76e360738ee19",
          "right_wrist": "abb51712d94f5459e750065f7d6700d308feb6007e6677f0915e686d02e8c39b"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "374ade935af5",
          "task": "robotwin/handover-block-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 284,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 284,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28280000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 284,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 284,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28280000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 284,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 284,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28280000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-02-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/02",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "02",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 23.6,
        "step_limit": 600,
        "steps": 590,
        "success": false,
        "valid": true
      },
      "steps": 590,
      "simulation_time_s": 23.6,
      "wall_time_s": 2309.686317,
      "timing": {
        "simulation_time_s": 23.6,
        "wall_time_s": 2309.686317,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "native_instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9470542913261929,
        "cache_read_input_tokens": 1396352,
        "cache_reported_input_tokens": 1474416,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1474416,
        "cached_input_tokens": 1396352,
        "cost_usd": null,
        "fresh_input_tokens": 78064,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1474416,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1396352,
        "known_fresh_input_tokens": 78064,
        "known_input_tokens": 1474416,
        "known_output_tokens": 8566,
        "known_reasoning_output_tokens": 2325,
        "output_tokens": 8566,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2325,
        "reasoning_reported_output_tokens": 8566,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 48,
          "cache_write_input_tokens": 48,
          "cache_write_reported_input_tokens": 48,
          "cached_input_tokens": 48,
          "fresh_input_tokens": 48,
          "input_tokens": 48,
          "output_tokens": 48,
          "reasoning_output_tokens": 48,
          "reasoning_reported_output_tokens": 48
        },
        "request_attempts": 48,
        "response_count": 48,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1482982,
        "uncached_input_tokens": 78064,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11822,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11953,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12001,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12290,
            "output_tokens": 252,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 17526,
            "output_tokens": 189,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 21268,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 22243,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22569,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22760,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23168,
            "output_tokens": 769,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 23958,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24164,
            "output_tokens": 53,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24279,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 24346,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25340,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25495,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 27599,
            "output_tokens": 109,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27735,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28500,
            "output_tokens": 80,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 29407,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29499,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30099,
            "output_tokens": 273,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30399,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31051,
            "output_tokens": 79,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31340,
            "output_tokens": 360,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32343,
            "output_tokens": 113,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32922,
            "output_tokens": 127,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33155,
            "output_tokens": 916,
            "reasoning_output_tokens": 139
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34092,
            "output_tokens": 131,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34739,
            "output_tokens": 50,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34851,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34930,
            "output_tokens": 130,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35385,
            "output_tokens": 142,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36291,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36955,
            "output_tokens": 630,
            "reasoning_output_tokens": 504
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37690,
            "output_tokens": 214,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38480,
            "output_tokens": 290,
            "reasoning_output_tokens": 155
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 38830,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 38922,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39379,
            "output_tokens": 143,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 40269,
            "output_tokens": 117,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40448,
            "output_tokens": 301,
            "reasoning_output_tokens": 194
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 41323,
            "output_tokens": 300,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41847,
            "output_tokens": 454,
            "reasoning_output_tokens": 302
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 43429,
            "output_tokens": 400,
            "reasoning_output_tokens": 257
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44964,
            "output_tokens": 141,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45869,
            "output_tokens": 600,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46492,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 57,
        "model_tool_calls_by_name": {
          "read": 16,
          "ls": 1,
          "bash": 1,
          "exec": 23,
          "devices": 2,
          "device": 1,
          "write": 3,
          "device_stop": 5,
          "grep": 1,
          "edit": 4
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 590 rosbridge control frames; 46 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/dual_shoes.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/workspace/.kinex/memos/dual_shoes.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/run_vla.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-02-20261113/workspace/.kinex/tools/run_vla.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 108,
        "observed_images": 21
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 46,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 237,
        "published_frames": 237,
        "decode_ok": true,
        "duration_s": 5.925,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "255fb8a14110a09515e4a38abe9caf86ef6c4644e3838570717c062637d350b8",
        "source_sha256": {
          "third_person": "4f60a0675078fc129f3863e257abfb0d0a7ba5111b89cccc53e3066c07d541a4",
          "left_wrist": "4462fdad570e7b3032448b3e6adfe197d45be9d686651f307891bad66742854e",
          "right_wrist": "d3c896a80c9747326e4392fda610b545e7892b23535667d287aa48442df0968c"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "6471dc92c016",
          "task": "robotwin/place-dual-shoes-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 238,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 237,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 238,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 237,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 238,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 237,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-03-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/03",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "03",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 2
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 54.800000000000004,
        "step_limit": 1700,
        "steps": 1370,
        "success": true,
        "valid": true
      },
      "steps": 1370,
      "simulation_time_s": 54.800000000000004,
      "wall_time_s": 3411.144817,
      "timing": {
        "simulation_time_s": 54.800000000000004,
        "wall_time_s": 3411.144817,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap in the dustbin the simple black garbage bin.",
      "native_instruction": "Place the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap in the dustbin the simple black garbage bin.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9782476620921082,
        "cache_read_input_tokens": 1722880,
        "cache_reported_input_tokens": 1761190,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1761190,
        "cached_input_tokens": 1722880,
        "cost_usd": null,
        "fresh_input_tokens": 38310,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1761190,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1722880,
        "known_fresh_input_tokens": 38310,
        "known_input_tokens": 1761190,
        "known_output_tokens": 9031,
        "known_reasoning_output_tokens": 1819,
        "output_tokens": 9031,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1819,
        "reasoning_reported_output_tokens": 9031,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 53,
          "cache_write_input_tokens": 53,
          "cache_write_reported_input_tokens": 53,
          "cached_input_tokens": 53,
          "fresh_input_tokens": 53,
          "input_tokens": 53,
          "output_tokens": 53,
          "reasoning_output_tokens": 53,
          "reasoning_reported_output_tokens": 53
        },
        "request_attempts": 53,
        "response_count": 53,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1770221,
        "uncached_input_tokens": 38310,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11838,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11878,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12011,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14994,
            "output_tokens": 217,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15104,
            "input_tokens": 17900,
            "output_tokens": 202,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 22502,
            "output_tokens": 96,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 24084,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24410,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 25826,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26016,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26424,
            "output_tokens": 607,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27053,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27276,
            "output_tokens": 49,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27387,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27454,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28442,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28579,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29435,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29525,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30089,
            "output_tokens": 535,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30652,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30789,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30879,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31477,
            "output_tokens": 109,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32243,
            "output_tokens": 99,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32552,
            "output_tokens": 394,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33627,
            "output_tokens": 886,
            "reasoning_output_tokens": 222
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34534,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34727,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34816,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35234,
            "output_tokens": 172,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35902,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36491,
            "output_tokens": 187,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 36797,
            "output_tokens": 259,
            "reasoning_output_tokens": 189
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37129,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37614,
            "output_tokens": 145,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38163,
            "output_tokens": 176,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38737,
            "output_tokens": 228,
            "reasoning_output_tokens": 135
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39368,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39858,
            "output_tokens": 332,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40903,
            "output_tokens": 237,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41723,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41917,
            "output_tokens": 257,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42573,
            "output_tokens": 178,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43422,
            "output_tokens": 181,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 44297,
            "output_tokens": 368,
            "reasoning_output_tokens": 189
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 45346,
            "output_tokens": 256,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 46376,
            "output_tokens": 231,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 47281,
            "output_tokens": 181,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47360,
            "input_tokens": 47608,
            "output_tokens": 59,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 47729,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 48199,
            "output_tokens": 880,
            "reasoning_output_tokens": 127
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49104,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 64,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 15,
          "find": 1,
          "exec": 26,
          "devices": 5,
          "device": 2,
          "write": 3,
          "device_stop": 5,
          "grep": 1,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 1370 rosbridge control frames; 73 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-bottles.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/workspace/.kinex/memos/robotwin-bottles.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_rollout.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-03-20261113/workspace/.kinex/tools/policy_rollout.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 120,
        "observed_images": 40
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 1700,
        "max_control_steps": 1700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 73,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 549,
        "published_frames": 549,
        "decode_ok": true,
        "duration_s": 13.725,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f0e41285e66b6acdb9c7d8b35599de6ece8b4daee6b6fccec2a49eb43e82ea45",
        "source_sha256": {
          "third_person": "5200da9cd4c693ecb4f4bbafcade458db2dc1d9cada707e9bfb6d5fad1cab349",
          "left_wrist": "38a36f9f752ee8a8d0795822551cb410f533c12ed60a398c2b2fe0bb73f912c0",
          "right_wrist": "9f85270df67d5df73ac3794fac60680bd19248830b219ea4bbebc1bd6be807d7"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "722a64b067e7",
          "task": "robotwin/put-bottles-dustbin-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 550,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 549,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 54800000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 550,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 549,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 54800000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 550,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 549,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 54800000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-04-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/04",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "04",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 19.92,
        "step_limit": 500,
        "steps": 498,
        "success": false,
        "valid": true
      },
      "steps": 498,
      "simulation_time_s": 19.92,
      "wall_time_s": 2373.707018,
      "timing": {
        "simulation_time_s": 19.92,
        "wall_time_s": 2373.707018,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "native_instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9626934274297821,
        "cache_read_input_tokens": 1154048,
        "cache_reported_input_tokens": 1198770,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1198770,
        "cached_input_tokens": 1154048,
        "cost_usd": null,
        "fresh_input_tokens": 44722,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1198770,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1154048,
        "known_fresh_input_tokens": 44722,
        "known_input_tokens": 1198770,
        "known_output_tokens": 5244,
        "known_reasoning_output_tokens": 448,
        "output_tokens": 5244,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 448,
        "reasoning_reported_output_tokens": 5244,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 42,
          "cache_write_input_tokens": 42,
          "cache_write_reported_input_tokens": 42,
          "cached_input_tokens": 42,
          "fresh_input_tokens": 42,
          "input_tokens": 42,
          "output_tokens": 42,
          "reasoning_output_tokens": 42,
          "reasoning_reported_output_tokens": 42
        },
        "request_attempts": 42,
        "response_count": 42,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1204014,
        "uncached_input_tokens": 44722,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11829,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11869,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12002,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12050,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14051,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13952,
            "input_tokens": 14340,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14208,
            "input_tokens": 14803,
            "output_tokens": 224,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 19582,
            "output_tokens": 114,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 21765,
            "output_tokens": 100,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 23025,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23351,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 24325,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24516,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24924,
            "output_tokens": 583,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25529,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25747,
            "output_tokens": 41,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25850,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 25980,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26869,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26959,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27376,
            "output_tokens": 72,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 29457,
            "output_tokens": 536,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30021,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 31132,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31266,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31356,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31736,
            "output_tokens": 105,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32884,
            "output_tokens": 86,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 34183,
            "output_tokens": 302,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 35031,
            "output_tokens": 90,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35331,
            "output_tokens": 752,
            "reasoning_output_tokens": 96
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36104,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36541,
            "output_tokens": 86,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36842,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37277,
            "output_tokens": 88,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 38393,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 39584,
            "output_tokens": 112,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40920,
            "output_tokens": 89,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 42106,
            "output_tokens": 104,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 43248,
            "output_tokens": 97,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 44076,
            "output_tokens": 441,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44540,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 49,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 15,
          "find": 1,
          "bash": 1,
          "exec": 18,
          "devices": 3,
          "device": 1,
          "write": 3,
          "device_stop": 3,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 498 rosbridge control frames; 106 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-policy.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/workspace/.kinex/memos/robotwin-policy.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_adjust.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/workspace/.kinex/tools/cartesian_adjust.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_rollout.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-04-20261113/workspace/.kinex/tools/policy_rollout.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 94,
        "observed_images": 33
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 106,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 200,
        "published_frames": 200,
        "decode_ok": true,
        "duration_s": 5.0,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "7945791e7f580644fe7de225425c23bc154d6d45e08ad4862fc964a841e2b6fe",
        "source_sha256": {
          "third_person": "615f93bfe22b26dae9e287b15b5ab0e5120ab0752dd527ad161e54c8165db089",
          "left_wrist": "848665761d7ea2591c744b70cd2a910f4f6f65ae32036fc86333ec85d43d8b99",
          "right_wrist": "14514d90d0bf245a6d5867f8686e9e2b60c96de80f41e4ca089d40446aa09034"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ba1592fedc45",
          "task": "robotwin/scan-object-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 201,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 200,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19920000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 201,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 200,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19920000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 201,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 200,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19920000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-05-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/05",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "05",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 40.28,
        "step_limit": 1200,
        "steps": 1007,
        "success": true,
        "valid": true
      },
      "steps": 1007,
      "simulation_time_s": 40.28,
      "wall_time_s": 3992.289429,
      "timing": {
        "simulation_time_s": 40.28,
        "wall_time_s": 3992.289429,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "native_instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9617191558197343,
        "cache_read_input_tokens": 2173568,
        "cache_reported_input_tokens": 2260086,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2260086,
        "cached_input_tokens": 2173568,
        "cost_usd": null,
        "fresh_input_tokens": 86518,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2260086,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2173568,
        "known_fresh_input_tokens": 86518,
        "known_input_tokens": 2260086,
        "known_output_tokens": 11267,
        "known_reasoning_output_tokens": 3577,
        "output_tokens": 11267,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3577,
        "reasoning_reported_output_tokens": 11267,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 64,
          "cache_write_input_tokens": 64,
          "cache_write_reported_input_tokens": 64,
          "cached_input_tokens": 64,
          "fresh_input_tokens": 64,
          "input_tokens": 64,
          "output_tokens": 64,
          "reasoning_output_tokens": 64,
          "reasoning_reported_output_tokens": 64
        },
        "request_attempts": 64,
        "response_count": 64,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2271353,
        "uncached_input_tokens": 86518,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11820,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11860,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11993,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12043,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13250,
            "output_tokens": 233,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13312,
            "input_tokens": 16172,
            "output_tokens": 212,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16256,
            "input_tokens": 20784,
            "output_tokens": 128,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 22642,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 22968,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23158,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23566,
            "output_tokens": 783,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 24372,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 25248,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25340,
            "output_tokens": 69,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25910,
            "output_tokens": 275,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26214,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26420,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 26509,
            "output_tokens": 55,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 28552,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28690,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28811,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28902,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29499,
            "output_tokens": 92,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30677,
            "output_tokens": 91,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31852,
            "output_tokens": 82,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31996,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 32794,
            "output_tokens": 96,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 32952,
            "output_tokens": 288,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33664,
            "output_tokens": 321,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34090,
            "output_tokens": 296,
            "reasoning_output_tokens": 254
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 35565,
            "output_tokens": 1206,
            "reasoning_output_tokens": 230
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 36794,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37493,
            "output_tokens": 85,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37607,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37750,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37839,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38257,
            "output_tokens": 142,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38797,
            "output_tokens": 61,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39029,
            "output_tokens": 166,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 39593,
            "output_tokens": 203,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 40193,
            "output_tokens": 104,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 41092,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41182,
            "output_tokens": 210,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 41421,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41729,
            "output_tokens": 491,
            "reasoning_output_tokens": 389
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42910,
            "output_tokens": 98,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43779,
            "output_tokens": 152,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 44621,
            "output_tokens": 133,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 44783,
            "output_tokens": 130,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45828,
            "output_tokens": 431,
            "reasoning_output_tokens": 298
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46657,
            "output_tokens": 282,
            "reasoning_output_tokens": 164
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 47336,
            "output_tokens": 378,
            "reasoning_output_tokens": 276
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 48112,
            "output_tokens": 365,
            "reasoning_output_tokens": 204
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 48873,
            "output_tokens": 242,
            "reasoning_output_tokens": 124
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49510,
            "output_tokens": 188,
            "reasoning_output_tokens": 86
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47488,
            "input_tokens": 50097,
            "output_tokens": 307,
            "reasoning_output_tokens": 162
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50304,
            "input_tokens": 50802,
            "output_tokens": 165,
            "reasoning_output_tokens": 102
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50816,
            "input_tokens": 51436,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49664,
            "input_tokens": 51933,
            "output_tokens": 196,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51456,
            "input_tokens": 52531,
            "output_tokens": 338,
            "reasoning_output_tokens": 236
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52736,
            "input_tokens": 53011,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52864,
            "input_tokens": 53102,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52992,
            "input_tokens": 53409,
            "output_tokens": 830,
            "reasoning_output_tokens": 131
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54016,
            "input_tokens": 54267,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 77,
        "model_tool_calls_by_name": {
          "ls": 5,
          "read": 18,
          "find": 1,
          "exec": 28,
          "devices": 6,
          "device": 3,
          "write": 3,
          "device_stop": 8,
          "edit": 5
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 1007 rosbridge control frames; 56 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-bowl-stacking.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/workspace/.kinex/memos/robotwin-bowl-stacking.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/run_vla.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-05-20261113/workspace/.kinex/tools/run_vla.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 144,
        "observed_images": 27
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 56,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 404,
        "published_frames": 404,
        "decode_ok": true,
        "duration_s": 10.1,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "96d94b950a678096615796c7bd0d6712c223b9cf55cbd875c15c6dc209b552a2",
        "source_sha256": {
          "third_person": "38130f4e2b36368d21ec1ae004e040a766902c38c541221a06f071cfbda615d4",
          "left_wrist": "2d6733ca7cfdc82c7bfda17a798a5386efac39feeb4d5f3721fd222e6c6addf1",
          "right_wrist": "5c72bd99b9aa9ecc09a1d136dc27ed6dc0ca3ecd3b439240e5fccd9be2553aa1"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "2508cfdc629f",
          "task": "robotwin/stack-bowls-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 404,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 404,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 40280000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 404,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 404,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 40280000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 404,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 404,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 40280000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-06-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/06",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "06",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 44.28,
        "step_limit": 1200,
        "steps": 1107,
        "success": true,
        "valid": true
      },
      "steps": 1107,
      "simulation_time_s": 44.28,
      "wall_time_s": 2563.746741,
      "timing": {
        "simulation_time_s": 44.28,
        "wall_time_s": 2563.746741,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "native_instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9746003942554574,
        "cache_read_input_tokens": 1725952,
        "cache_reported_input_tokens": 1770933,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1770933,
        "cached_input_tokens": 1725952,
        "cost_usd": null,
        "fresh_input_tokens": 44981,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1770933,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1725952,
        "known_fresh_input_tokens": 44981,
        "known_input_tokens": 1770933,
        "known_output_tokens": 8965,
        "known_reasoning_output_tokens": 1756,
        "output_tokens": 8965,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1756,
        "reasoning_reported_output_tokens": 8965,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 55,
          "cache_write_input_tokens": 55,
          "cache_write_reported_input_tokens": 55,
          "cached_input_tokens": 55,
          "fresh_input_tokens": 55,
          "input_tokens": 55,
          "output_tokens": 55,
          "reasoning_output_tokens": 55,
          "reasoning_reported_output_tokens": 55
        },
        "request_attempts": 55,
        "response_count": 55,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1779898,
        "uncached_input_tokens": 44981,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11820,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11860,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11993,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12333,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 15936,
            "output_tokens": 248,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 20985,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21311,
            "output_tokens": 132,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 22887,
            "output_tokens": 573,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23482,
            "output_tokens": 65,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24337,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24429,
            "output_tokens": 363,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24820,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25031,
            "output_tokens": 41,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25134,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25252,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25378,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25469,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26070,
            "output_tokens": 85,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26906,
            "output_tokens": 80,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27732,
            "output_tokens": 176,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 29083,
            "output_tokens": 582,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29686,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30312,
            "output_tokens": 167,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30506,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31131,
            "output_tokens": 132,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 32906,
            "output_tokens": 1100,
            "reasoning_output_tokens": 195
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34027,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34183,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34274,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34691,
            "output_tokens": 122,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35327,
            "output_tokens": 41,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35430,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35508,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35870,
            "output_tokens": 92,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36248,
            "output_tokens": 209,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 37153,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37246,
            "output_tokens": 233,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 38039,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38128,
            "output_tokens": 94,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38618,
            "output_tokens": 153,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39360,
            "output_tokens": 423,
            "reasoning_output_tokens": 273
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40231,
            "output_tokens": 125,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40692,
            "output_tokens": 127,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 41179,
            "output_tokens": 64,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41831,
            "output_tokens": 270,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42546,
            "output_tokens": 194,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43186,
            "output_tokens": 282,
            "reasoning_output_tokens": 136
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43905,
            "output_tokens": 224,
            "reasoning_output_tokens": 117
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44470,
            "output_tokens": 161,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44773,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 44867,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45060,
            "output_tokens": 326,
            "reasoning_output_tokens": 141
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45413,
            "output_tokens": 101,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45542,
            "output_tokens": 783,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46347,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 70,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 17,
          "bash": 1,
          "exec": 24,
          "devices": 5,
          "device": 3,
          "write": 4,
          "device_stop": 8,
          "edit": 6,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 1107 rosbridge control frames; 66 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack_blocks.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/workspace/.kinex/memos/stack_blocks.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/block_observation.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/workspace/.kinex/tools/block_observation.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_rollout.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-06-20261113/workspace/.kinex/tools/policy_rollout.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 128,
        "observed_images": 26
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 66,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 444,
        "published_frames": 444,
        "decode_ok": true,
        "duration_s": 11.1,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "3ea000a804d5f3e9c4f632fe093ba90c293acdc4f49deb029af2e40ed88e0d53",
        "source_sha256": {
          "third_person": "eed52bfbc2ea52e967d99bc75ea79cd2cd1fd5f49651f78a1d093c7ee47f2c5f",
          "left_wrist": "35d9841ec5540c35075f63e9b727a3974037a40bfa98815ce17af378d6126a95",
          "right_wrist": "ac4e5c78c4b6650f7ad49831b9a698234cbe84d2e2f2c11b73ca6770a8ce2c9b"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "df9902873d13",
          "task": "robotwin/stack-blocks-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 444,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 444,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 44280000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 444,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 444,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 44280000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 444,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 444,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 44280000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-07-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/07",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "07",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 34.88,
        "step_limit": 900,
        "steps": 872,
        "success": false,
        "valid": true
      },
      "steps": 872,
      "simulation_time_s": 34.88,
      "wall_time_s": 4328.485652,
      "timing": {
        "simulation_time_s": 34.88,
        "wall_time_s": 4328.485652,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the cylindrical mug with handle on the table, rotate it, set it down in the middle, then hang it onto the rack with straight and slanted rods.",
      "native_instruction": "Grab the cylindrical mug with handle on the table, rotate it, set it down in the middle, then hang it onto the rack with straight and slanted rods.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9854647120470935,
        "cache_read_input_tokens": 3486720,
        "cache_reported_input_tokens": 3538148,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 3538148,
        "cached_input_tokens": 3486720,
        "cost_usd": null,
        "fresh_input_tokens": 51428,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 3538148,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 3486720,
        "known_fresh_input_tokens": 51428,
        "known_input_tokens": 3538148,
        "known_output_tokens": 14606,
        "known_reasoning_output_tokens": 5348,
        "output_tokens": 14606,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 5348,
        "reasoning_reported_output_tokens": 14606,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 91,
          "cache_write_input_tokens": 91,
          "cache_write_reported_input_tokens": 91,
          "cached_input_tokens": 91,
          "fresh_input_tokens": 91,
          "input_tokens": 91,
          "output_tokens": 91,
          "reasoning_output_tokens": 91,
          "reasoning_reported_output_tokens": 91
        },
        "request_attempts": 91,
        "response_count": 91,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 3552754,
        "uncached_input_tokens": 51428,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11837,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11968,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12016,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13957,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14080,
            "input_tokens": 19534,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 20003,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 20573,
            "output_tokens": 21,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20608,
            "output_tokens": 158,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 20984,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 22370,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22561,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22969,
            "output_tokens": 559,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23548,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24400,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24488,
            "output_tokens": 63,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25054,
            "output_tokens": 228,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25308,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25495,
            "output_tokens": 108,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25629,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 26190,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26290,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26380,
            "output_tokens": 55,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26604,
            "output_tokens": 86,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27206,
            "output_tokens": 61,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27314,
            "output_tokens": 66,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27905,
            "output_tokens": 219,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28861,
            "output_tokens": 186,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29633,
            "output_tokens": 267,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30350,
            "output_tokens": 71,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30887,
            "output_tokens": 829,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31737,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31854,
            "output_tokens": 113,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32589,
            "output_tokens": 63,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32679,
            "output_tokens": 113,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33584,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33674,
            "output_tokens": 76,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34090,
            "output_tokens": 138,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34255,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34683,
            "output_tokens": 96,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35118,
            "output_tokens": 99,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35557,
            "output_tokens": 420,
            "reasoning_output_tokens": 190
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36304,
            "output_tokens": 245,
            "reasoning_output_tokens": 131
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36891,
            "output_tokens": 148,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37734,
            "output_tokens": 78,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 37874,
            "output_tokens": 123,
            "reasoning_output_tokens": 86
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38055,
            "output_tokens": 415,
            "reasoning_output_tokens": 325
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38816,
            "output_tokens": 63,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39222,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39644,
            "output_tokens": 214,
            "reasoning_output_tokens": 139
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 39930,
            "output_tokens": 107,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40728,
            "output_tokens": 97,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41784,
            "output_tokens": 125,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42248,
            "output_tokens": 116,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42705,
            "output_tokens": 589,
            "reasoning_output_tokens": 473
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43639,
            "output_tokens": 317,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 44314,
            "output_tokens": 330,
            "reasoning_output_tokens": 184
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44986,
            "output_tokens": 111,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45437,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45860,
            "output_tokens": 136,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46333,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46728,
            "output_tokens": 140,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 47207,
            "output_tokens": 103,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 47372,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47734,
            "output_tokens": 607,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48128,
            "input_tokens": 48684,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48512,
            "input_tokens": 48877,
            "output_tokens": 236,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49371,
            "output_tokens": 173,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49408,
            "input_tokens": 49885,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49792,
            "input_tokens": 50280,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50176,
            "input_tokens": 50705,
            "output_tokens": 95,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50688,
            "input_tokens": 50969,
            "output_tokens": 490,
            "reasoning_output_tokens": 414
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51328,
            "input_tokens": 51525,
            "output_tokens": 104,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51456,
            "input_tokens": 51964,
            "output_tokens": 56,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51840,
            "input_tokens": 52189,
            "output_tokens": 479,
            "reasoning_output_tokens": 255
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52480,
            "input_tokens": 53027,
            "output_tokens": 398,
            "reasoning_output_tokens": 294
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53248,
            "input_tokens": 54116,
            "output_tokens": 107,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54016,
            "input_tokens": 54742,
            "output_tokens": 64,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54656,
            "input_tokens": 56450,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56320,
            "input_tokens": 56554,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56448,
            "input_tokens": 57380,
            "output_tokens": 116,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57344,
            "input_tokens": 57835,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57600,
            "input_tokens": 58027,
            "output_tokens": 317,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58240,
            "input_tokens": 58629,
            "output_tokens": 222,
            "reasoning_output_tokens": 118
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58752,
            "input_tokens": 59190,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59136,
            "input_tokens": 59383,
            "output_tokens": 330,
            "reasoning_output_tokens": 224
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59648,
            "input_tokens": 60053,
            "output_tokens": 82,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60032,
            "input_tokens": 60197,
            "output_tokens": 386,
            "reasoning_output_tokens": 274
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60416,
            "input_tokens": 60922,
            "output_tokens": 308,
            "reasoning_output_tokens": 121
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61056,
            "input_tokens": 61256,
            "output_tokens": 188,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61312,
            "input_tokens": 61471,
            "output_tokens": 686,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61952,
            "input_tokens": 62180,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 100,
        "model_tool_calls_by_name": {
          "read": 18,
          "ls": 2,
          "find": 1,
          "exec": 48,
          "devices": 4,
          "device": 8,
          "write": 3,
          "device_stop": 6,
          "edit": 7,
          "relative_pose": 2,
          "grep": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 872 rosbridge control frames; 86 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/hanging-mug.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/workspace/.kinex/memos/hanging-mug.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/move_tcp.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/workspace/.kinex/tools/move_tcp.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/run_policy.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-07-20261113/workspace/.kinex/tools/run_policy.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 194,
        "observed_images": 41
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 86,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 350,
        "published_frames": 350,
        "decode_ok": true,
        "duration_s": 8.75,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f48b751788791cb3160d3193d498f28c3fc2dedbbaf3e217465a5c859a128f0d",
        "source_sha256": {
          "third_person": "4a93f26a530982373296d016165cc0d76805c4ab038838dc86e11b0236ca8dd7",
          "left_wrist": "5eeff6fab2393b46b1535bf3f0730d8cb2506d791c8d495c04b4fcad4cadad1e",
          "right_wrist": "c023c7a6fee4720b6f525fe694a2ea745e281f0a01b063c4c4bb75a41818afb2"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "7a74e1adfb12",
          "task": "robotwin/hanging-mug-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 350,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 350,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 34880000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 350,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 350,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 34880000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 350,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 350,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 34880000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-08-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/08",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "08",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 24.88,
        "step_limit": 700,
        "steps": 622,
        "success": false,
        "valid": true
      },
      "steps": 622,
      "simulation_time_s": 24.88,
      "wall_time_s": 3373.666005,
      "timing": {
        "simulation_time_s": 24.88,
        "wall_time_s": 3373.666005,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Open the cabinet with smooth finish's drawer and set the blue soap with printed patterns inside with the other arm.",
      "native_instruction": "Open the cabinet with smooth finish's drawer and set the blue soap with printed patterns inside with the other arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9726558837554148,
        "cache_read_input_tokens": 2417152,
        "cache_reported_input_tokens": 2485105,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2485105,
        "cached_input_tokens": 2417152,
        "cost_usd": null,
        "fresh_input_tokens": 67953,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2485105,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2417152,
        "known_fresh_input_tokens": 67953,
        "known_input_tokens": 2485105,
        "known_output_tokens": 13577,
        "known_reasoning_output_tokens": 5375,
        "output_tokens": 13577,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 5375,
        "reasoning_reported_output_tokens": 13577,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 67,
          "cache_write_input_tokens": 67,
          "cache_write_reported_input_tokens": 67,
          "cached_input_tokens": 67,
          "fresh_input_tokens": 67,
          "input_tokens": 67,
          "output_tokens": 67,
          "reasoning_output_tokens": 67,
          "reasoning_reported_output_tokens": 67
        },
        "request_attempts": 67,
        "response_count": 67,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2498682,
        "uncached_input_tokens": 67953,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11826,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11866,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11999,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12047,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 12336,
            "output_tokens": 252,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 17572,
            "output_tokens": 165,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 19851,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 22687,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23661,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 23987,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24177,
            "output_tokens": 84,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24630,
            "output_tokens": 747,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25398,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26280,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26372,
            "output_tokens": 132,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26531,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26743,
            "output_tokens": 123,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26893,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27020,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27110,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27708,
            "output_tokens": 107,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28607,
            "output_tokens": 86,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29487,
            "output_tokens": 105,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30371,
            "output_tokens": 257,
            "reasoning_output_tokens": 156
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30833,
            "output_tokens": 255,
            "reasoning_output_tokens": 142
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31928,
            "output_tokens": 352,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32857,
            "output_tokens": 76,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 32995,
            "output_tokens": 240,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33606,
            "output_tokens": 250,
            "reasoning_output_tokens": 208
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 35035,
            "output_tokens": 916,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 35972,
            "output_tokens": 181,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36213,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36304,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36705,
            "output_tokens": 239,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37153,
            "output_tokens": 257,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 38254,
            "output_tokens": 113,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 39103,
            "output_tokens": 108,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39233,
            "output_tokens": 74,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 40042,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40908,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41001,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41239,
            "output_tokens": 281,
            "reasoning_output_tokens": 162
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41757,
            "output_tokens": 131,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 41915,
            "output_tokens": 473,
            "reasoning_output_tokens": 222
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42627,
            "output_tokens": 272,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43186,
            "output_tokens": 436,
            "reasoning_output_tokens": 352
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 44307,
            "output_tokens": 107,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 45204,
            "output_tokens": 92,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45621,
            "output_tokens": 261,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46314,
            "output_tokens": 511,
            "reasoning_output_tokens": 426
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 47616,
            "output_tokens": 451,
            "reasoning_output_tokens": 302
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47872,
            "input_tokens": 48201,
            "output_tokens": 195,
            "reasoning_output_tokens": 113
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 49184,
            "output_tokens": 122,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49280,
            "input_tokens": 50095,
            "output_tokens": 358,
            "reasoning_output_tokens": 273
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50304,
            "input_tokens": 51241,
            "output_tokens": 301,
            "reasoning_output_tokens": 255
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51328,
            "input_tokens": 52276,
            "output_tokens": 697,
            "reasoning_output_tokens": 613
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52864,
            "input_tokens": 53759,
            "output_tokens": 73,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53632,
            "input_tokens": 54565,
            "output_tokens": 355,
            "reasoning_output_tokens": 226
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54784,
            "input_tokens": 55381,
            "output_tokens": 313,
            "reasoning_output_tokens": 183
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55552,
            "input_tokens": 56155,
            "output_tokens": 269,
            "reasoning_output_tokens": 139
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56192,
            "input_tokens": 56883,
            "output_tokens": 132,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56832,
            "input_tokens": 57042,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56960,
            "input_tokens": 57953,
            "output_tokens": 631,
            "reasoning_output_tokens": 547
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58496,
            "input_tokens": 59270,
            "output_tokens": 182,
            "reasoning_output_tokens": 148
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59392,
            "input_tokens": 59514,
            "output_tokens": 671,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60032,
            "input_tokens": 60210,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60160,
            "input_tokens": 60289,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 78,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 17,
          "bash": 1,
          "exec": 36,
          "devices": 5,
          "device": 2,
          "write": 3,
          "device_stop": 7,
          "edit": 5
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 622 rosbridge control frames; 50 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-cabinet.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/workspace/.kinex/memos/robotwin-cabinet.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vla_batch.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-08-20261113/workspace/.kinex/tools/vla_batch.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 148,
        "observed_images": 54
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 50,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 250,
        "published_frames": 250,
        "decode_ok": true,
        "duration_s": 6.25,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "06fc587308937ff2929b0d3e75618899b5dfe5711c7f97ec682745a1a403058f",
        "source_sha256": {
          "third_person": "b21e3090d5112952bb375f415696d9bfa20167fd36bb12e0ccbd228c2ba19a27",
          "left_wrist": "5df8f01672acb0e885d921232d7274601540adf0a487b2fe0217649c79b6d7f0",
          "right_wrist": "2f872faeb62df7eb197b5af02e320141a07acf2df85cb1cdeba23e3ea7937848"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "52e07e08ea23",
          "task": "robotwin/put-object-cabinet-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 250,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 250,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24880000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 250,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 250,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24880000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 250,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 250,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24880000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-09-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/09",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "09",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 35.92,
        "step_limit": 1200,
        "steps": 898,
        "success": true,
        "valid": true
      },
      "steps": 898,
      "simulation_time_s": 35.92,
      "wall_time_s": 2617.464628,
      "timing": {
        "simulation_time_s": 35.92,
        "wall_time_s": 2617.464628,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move large block, medium block, and small block to the center table.",
      "native_instruction": "Move large block, medium block, and small block to the center table.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9784997611084567,
        "cache_read_input_tokens": 1652736,
        "cache_reported_input_tokens": 1689051,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1689051,
        "cached_input_tokens": 1652736,
        "cost_usd": null,
        "fresh_input_tokens": 36315,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1689051,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1652736,
        "known_fresh_input_tokens": 36315,
        "known_input_tokens": 1689051,
        "known_output_tokens": 6900,
        "known_reasoning_output_tokens": 868,
        "output_tokens": 6900,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 868,
        "reasoning_reported_output_tokens": 6900,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 51,
          "cache_write_input_tokens": 51,
          "cache_write_reported_input_tokens": 51,
          "cached_input_tokens": 51,
          "fresh_input_tokens": 51,
          "input_tokens": 51,
          "output_tokens": 51,
          "reasoning_output_tokens": 51,
          "reasoning_reported_output_tokens": 51
        },
        "request_attempts": 51,
        "response_count": 51,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1695951,
        "uncached_input_tokens": 36315,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11859,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11992,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15619,
            "output_tokens": 244,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 19297,
            "output_tokens": 231,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 23632,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24921,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25247,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25439,
            "output_tokens": 86,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25894,
            "output_tokens": 649,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26567,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26758,
            "output_tokens": 45,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26865,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26992,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27856,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 27946,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28364,
            "output_tokens": 351,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28745,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29332,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29914,
            "output_tokens": 112,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30236,
            "output_tokens": 279,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30934,
            "output_tokens": 245,
            "reasoning_output_tokens": 116
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 33515,
            "output_tokens": 1114,
            "reasoning_output_tokens": 257
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34652,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35086,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35515,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35938,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36381,
            "output_tokens": 119,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36857,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37271,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37709,
            "output_tokens": 158,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37927,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38018,
            "output_tokens": 75,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38466,
            "output_tokens": 114,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38930,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39343,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 40213,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40304,
            "output_tokens": 140,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40473,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40712,
            "output_tokens": 102,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41166,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41581,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 42028,
            "output_tokens": 155,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42539,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42954,
            "output_tokens": 105,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 43413,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43860,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 44061,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44156,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44496,
            "output_tokens": 738,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45259,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 64,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 18,
          "find": 1,
          "exec": 25,
          "devices": 5,
          "device": 1,
          "write": 3,
          "device_stop": 5,
          "edit": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 898 rosbridge control frames; 59 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/blocks-ranking-size.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/workspace/.kinex/memos/blocks-ranking-size.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_rollout.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-09-20261113/workspace/.kinex/tools/policy_rollout.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 118,
        "observed_images": 23
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 59,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 360,
        "published_frames": 360,
        "decode_ok": true,
        "duration_s": 9.0,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "4c5d5a6a3fbf38e0002a9da16b7e903341f94b34c2893ec433a0f64711628c4a",
        "source_sha256": {
          "third_person": "a6f83ba1d8d825ececc124f8f4e466728eed95ecf96c298540a03cf9b8969197",
          "left_wrist": "48a33463618cb5b512df35778cdb52e62fbacc9f38b8c0ff8f29fc1061fef7e5",
          "right_wrist": "b3eb39681c5e2b44713e6fcec9f044ccf6b1868462403476b02142db0fdd33b1"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "8f1bb4218cb1",
          "task": "robotwin/blocks-ranking-size-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 361,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 360,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 35920000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 361,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 360,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 35920000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 361,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 360,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 35920000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-10-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/10",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "10",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 10.44,
        "step_limit": 400,
        "steps": 261,
        "success": true,
        "valid": true
      },
      "steps": 261,
      "simulation_time_s": 10.44,
      "wall_time_s": 1318.444814,
      "timing": {
        "simulation_time_s": 10.44,
        "wall_time_s": 1318.444814,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Hold the dark gray cooking kitchenpot firmly and lift it",
      "native_instruction": "Hold the dark gray cooking kitchenpot firmly and lift it",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9635869025299144,
        "cache_read_input_tokens": 911744,
        "cache_reported_input_tokens": 946198,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 946198,
        "cached_input_tokens": 911744,
        "cost_usd": null,
        "fresh_input_tokens": 34454,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 946198,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 911744,
        "known_fresh_input_tokens": 34454,
        "known_input_tokens": 946198,
        "known_output_tokens": 4960,
        "known_reasoning_output_tokens": 558,
        "output_tokens": 4960,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 558,
        "reasoning_reported_output_tokens": 4960,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 951158,
        "uncached_input_tokens": 34454,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11810,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11850,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11983,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12033,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13093,
            "output_tokens": 264,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13184,
            "input_tokens": 16803,
            "output_tokens": 181,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 20627,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 20953,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21927,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22118,
            "output_tokens": 568,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22709,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23546,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23633,
            "output_tokens": 340,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24002,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24208,
            "output_tokens": 51,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 25895,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26023,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 27888,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 30466,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 33018,
            "output_tokens": 160,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 34545,
            "output_tokens": 322,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35608,
            "output_tokens": 139,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35945,
            "output_tokens": 165,
            "reasoning_output_tokens": 104
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 37897,
            "output_tokens": 85,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 39161,
            "output_tokens": 907,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40091,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40897,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41704,
            "output_tokens": 68,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42494,
            "output_tokens": 113,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42749,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 42839,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43533,
            "output_tokens": 593,
            "reasoning_output_tokens": 155
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44150,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 46,
        "model_tool_calls_by_name": {
          "ls": 5,
          "read": 17,
          "find": 1,
          "devices": 2,
          "exec": 13,
          "device": 1,
          "write": 3,
          "device_stop": 2,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 261 rosbridge control frames; 40 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/lift-pot.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/workspace/.kinex/memos/lift-pot.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_tcp.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/workspace/.kinex/tools/dual_tcp.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vla_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-10-20261113/workspace/.kinex/tools/vla_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 82,
        "observed_images": 23
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 40,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 106,
        "published_frames": 106,
        "decode_ok": true,
        "duration_s": 2.65,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "9894b5b50a75f3bec7c5544fb026732f89458ec34dc8b15ffc68c539b5f44d46",
        "source_sha256": {
          "third_person": "8011a8bdcafde1efa7ca92a5fe081334e6aeb4c0331d67b52c1302ed6b68ed90",
          "left_wrist": "3c36db393aca1c093b7daef05f2dfa1a720849cd0f067f8ceecb5d7ef28baecd",
          "right_wrist": "17e9ef38ae63b7720f4874bcb50a68dc2b56133027783b54e2af5cb8fbf5751c"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "a5063ca12ad6",
          "task": "robotwin/lift-pot-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 106,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 106,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10440000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 106,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 106,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10440000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 106,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 106,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10440000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-11-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/11",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "11",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 2
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 14.32,
        "step_limit": 400,
        "steps": 358,
        "success": false,
        "valid": true
      },
      "steps": 358,
      "simulation_time_s": 14.32,
      "wall_time_s": 1159.244322,
      "timing": {
        "simulation_time_s": 14.32,
        "wall_time_s": 1159.244322,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the green plastic bottle with ridged bottom carefully using the right arm",
      "native_instruction": "Pick up the green plastic bottle with ridged bottom carefully using the right arm",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9736985381292328,
        "cache_read_input_tokens": 1148160,
        "cache_reported_input_tokens": 1179174,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1179174,
        "cached_input_tokens": 1148160,
        "cost_usd": null,
        "fresh_input_tokens": 31014,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1179174,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1148160,
        "known_fresh_input_tokens": 31014,
        "known_input_tokens": 1179174,
        "known_output_tokens": 5667,
        "known_reasoning_output_tokens": 573,
        "output_tokens": 5667,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 573,
        "reasoning_reported_output_tokens": 5667,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 41,
          "cache_write_input_tokens": 41,
          "cache_write_reported_input_tokens": 41,
          "cached_input_tokens": 41,
          "fresh_input_tokens": 41,
          "input_tokens": 41,
          "output_tokens": 41,
          "reasoning_output_tokens": 41,
          "reasoning_reported_output_tokens": 41
        },
        "request_attempts": 41,
        "response_count": 41,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1184841,
        "uncached_input_tokens": 31014,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11951,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12003,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12292,
            "output_tokens": 191,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 15900,
            "output_tokens": 197,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 19733,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20948,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21274,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21465,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 22439,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22847,
            "output_tokens": 599,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23467,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23674,
            "output_tokens": 110,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 23811,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 25136,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 27172,
            "output_tokens": 78,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 29219,
            "output_tokens": 100,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29346,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29952,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30788,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30877,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31363,
            "output_tokens": 317,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31707,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32319,
            "output_tokens": 85,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32956,
            "output_tokens": 64,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33230,
            "output_tokens": 270,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33967,
            "output_tokens": 108,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34541,
            "output_tokens": 141,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34794,
            "output_tokens": 127,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 34983,
            "output_tokens": 660,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35664,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35825,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36198,
            "output_tokens": 109,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36589,
            "output_tokens": 144,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37012,
            "output_tokens": 184,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37396,
            "output_tokens": 128,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37805,
            "output_tokens": 147,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38637,
            "output_tokens": 61,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 38720,
            "output_tokens": 336,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39437,
            "output_tokens": 458,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 39919,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 49,
        "model_tool_calls_by_name": {
          "read": 15,
          "ls": 1,
          "bash": 1,
          "exec": 20,
          "devices": 2,
          "device": 1,
          "write": 3,
          "edit": 3,
          "device_stop": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 358 rosbridge control frames; 60 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin_bottle.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/workspace/.kinex/memos/robotwin_bottle.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_steps.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/workspace/.kinex/tools/policy_steps.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/right_tcp_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-11-20261113/workspace/.kinex/tools/right_tcp_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 93,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 60,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 144,
        "published_frames": 144,
        "decode_ok": true,
        "duration_s": 3.6,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "931d8c122d14be35d4256cc923a91b9065fc85e47e7618175719577a502f4dd7",
        "source_sha256": {
          "third_person": "13a4869fafcc15367efa2d1c3d140e7fce99fab5e6a23f0cd809bc420a1384a8",
          "left_wrist": "63a568ec40005feebcf1acacfb7b00d498b57d54d162bf90005308308e47a72b",
          "right_wrist": "273885b4322aa2a90804c0c6815c91589d8aef6e3d90a8ba2e15b29118c04121"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "cb4beaebd65d",
          "task": "robotwin/adjust-bottle-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 145,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 144,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14320000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 145,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 144,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14320000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 145,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 144,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14320000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-12-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/12",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "12",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.4,
        "step_limit": 400,
        "steps": 310,
        "success": true,
        "valid": true
      },
      "steps": 310,
      "simulation_time_s": 12.4,
      "wall_time_s": 1457.168667,
      "timing": {
        "simulation_time_s": 12.4,
        "wall_time_s": 1457.168667,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the hammer with black handle and hit the block",
      "native_instruction": "Grab the hammer with black handle and hit the block",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.938885085324523,
        "cache_read_input_tokens": 991488,
        "cache_reported_input_tokens": 1056027,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1056027,
        "cached_input_tokens": 991488,
        "cost_usd": null,
        "fresh_input_tokens": 64539,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1056027,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 991488,
        "known_fresh_input_tokens": 64539,
        "known_input_tokens": 1056027,
        "known_output_tokens": 5251,
        "known_reasoning_output_tokens": 972,
        "output_tokens": 5251,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 972,
        "reasoning_reported_output_tokens": 5251,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 40,
          "cache_write_input_tokens": 40,
          "cache_write_reported_input_tokens": 40,
          "cached_input_tokens": 40,
          "fresh_input_tokens": 40,
          "input_tokens": 40,
          "output_tokens": 40,
          "reasoning_output_tokens": 40,
          "reasoning_reported_output_tokens": 40
        },
        "request_attempts": 40,
        "response_count": 40,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1061278,
        "uncached_input_tokens": 64539,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11814,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11854,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11987,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 12037,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13975,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13824,
            "input_tokens": 14061,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13952,
            "input_tokens": 14878,
            "output_tokens": 223,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14976,
            "input_tokens": 19303,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 22694,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23020,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 24436,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24627,
            "output_tokens": 60,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25056,
            "output_tokens": 470,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25547,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25747,
            "output_tokens": 134,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25908,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26298,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26412,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26501,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26915,
            "output_tokens": 70,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27324,
            "output_tokens": 72,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27732,
            "output_tokens": 69,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28011,
            "output_tokens": 260,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28567,
            "output_tokens": 84,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29117,
            "output_tokens": 704,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29842,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30313,
            "output_tokens": 100,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30805,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31241,
            "output_tokens": 65,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31992,
            "output_tokens": 94,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32476,
            "output_tokens": 149,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33320,
            "output_tokens": 306,
            "reasoning_output_tokens": 211
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33692,
            "output_tokens": 121,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 34201,
            "output_tokens": 68,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34593,
            "output_tokens": 246,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35031,
            "output_tokens": 275,
            "reasoning_output_tokens": 174
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35695,
            "output_tokens": 62,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35899,
            "output_tokens": 80,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36242,
            "output_tokens": 598,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36864,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 48,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "find": 1,
          "bash": 1,
          "exec": 18,
          "devices": 4,
          "device": 1,
          "write": 3,
          "edit": 1,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 310 rosbridge control frames; 40 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/beat-block-hammer.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/workspace/.kinex/memos/beat-block-hammer.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-12-20261113/workspace/.kinex/tools/policy_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 91,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 40,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 125,
        "published_frames": 125,
        "decode_ok": true,
        "duration_s": 3.125,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "4d016e217a39cf9e0b4f19d6de9af75b34accf0dc89bcf22bb1328d795458422",
        "source_sha256": {
          "third_person": "c8a7f7a25c3a1116a27ca1bd38e74afbabb9fe9417bacb7060fbc478b20e715d",
          "left_wrist": "efe0192ab4c2f389510eec0cece482d598c4c82676aa50f29ceff09693a337e2",
          "right_wrist": "fc283d193600362909691b488629b7248f36b8d471d07cee4a6728d35b6a29ca"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "63ae55f30c67",
          "task": "robotwin/beat-block-hammer-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 126,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 125,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12400000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 126,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 125,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12400000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 126,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 125,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12400000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-13-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/13",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "13",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 39.64,
        "step_limit": 1200,
        "steps": 991,
        "success": true,
        "valid": true
      },
      "steps": 991,
      "simulation_time_s": 39.64,
      "wall_time_s": 2785.07311,
      "timing": {
        "simulation_time_s": 39.64,
        "wall_time_s": 2785.07311,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set red block on the left, then green block in the center, and finally blue block on the right.",
      "native_instruction": "Set red block on the left, then green block in the center, and finally blue block on the right.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.978178690630183,
        "cache_read_input_tokens": 1715072,
        "cache_reported_input_tokens": 1753332,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1753332,
        "cached_input_tokens": 1715072,
        "cost_usd": null,
        "fresh_input_tokens": 38260,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1753332,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1715072,
        "known_fresh_input_tokens": 38260,
        "known_input_tokens": 1753332,
        "known_output_tokens": 7908,
        "known_reasoning_output_tokens": 1447,
        "output_tokens": 7908,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1447,
        "reasoning_reported_output_tokens": 7908,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 51,
          "cache_write_input_tokens": 51,
          "cache_write_reported_input_tokens": 51,
          "cached_input_tokens": 51,
          "fresh_input_tokens": 51,
          "input_tokens": 51,
          "output_tokens": 51,
          "reasoning_output_tokens": 51,
          "reasoning_reported_output_tokens": 51
        },
        "request_attempts": 51,
        "response_count": 51,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1761240,
        "uncached_input_tokens": 38260,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11826,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11866,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11999,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 16502,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 16791,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 17254,
            "output_tokens": 224,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 22033,
            "output_tokens": 197,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 25783,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26109,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26298,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 27715,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28123,
            "output_tokens": 626,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28771,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29619,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29708,
            "output_tokens": 65,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30124,
            "output_tokens": 294,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30446,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30659,
            "output_tokens": 55,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 32143,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32268,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32818,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32946,
            "output_tokens": 42,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33050,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33645,
            "output_tokens": 129,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34260,
            "output_tokens": 53,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34375,
            "output_tokens": 111,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 35328,
            "output_tokens": 429,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36292,
            "output_tokens": 334,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37333,
            "output_tokens": 282,
            "reasoning_output_tokens": 240
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38794,
            "output_tokens": 1027,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 39842,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40437,
            "output_tokens": 50,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40509,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40592,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 41007,
            "output_tokens": 110,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41358,
            "output_tokens": 139,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41806,
            "output_tokens": 225,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42453,
            "output_tokens": 225,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 43105,
            "output_tokens": 186,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43351,
            "output_tokens": 58,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43471,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43962,
            "output_tokens": 184,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44448,
            "output_tokens": 255,
            "reasoning_output_tokens": 183
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 45026,
            "output_tokens": 216,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45492,
            "output_tokens": 190,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 46104,
            "output_tokens": 162,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46408,
            "output_tokens": 50,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46520,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 46971,
            "output_tokens": 766,
            "reasoning_output_tokens": 182
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 47761,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 47831,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 61,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 17,
          "find": 1,
          "bash": 1,
          "exec": 19,
          "devices": 7,
          "device": 2,
          "write": 3,
          "device_stop": 6,
          "edit": 4
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 991 rosbridge control frames; 75 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-block-ranking.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/workspace/.kinex/memos/robotwin-block-ranking.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vla_rollout.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-13-20261113/workspace/.kinex/tools/vla_rollout.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 115,
        "observed_images": 18
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 75,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 398,
        "published_frames": 398,
        "decode_ok": true,
        "duration_s": 9.95,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "28dac32be330259b7341850277c4bd087cad0f6d644e4c1132989b7d1040fde5",
        "source_sha256": {
          "third_person": "9f231e96272ea67dcb7f3c3f3c6254346d6ccd74030d8de5b5e85ceab768312b",
          "left_wrist": "f1bdde11236bc51e82a106b194027e9187090792cb866c48932e9e6a24413996",
          "right_wrist": "3f00f8a27580e012714cd3261693329117863b6bda71351439385a438f6a4b94"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "6e8fbf237667",
          "task": "robotwin/blocks-ranking-rgb-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 398,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 398,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 39640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 398,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 398,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 39640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 398,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 398,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 39640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-14-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/14",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "14",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 8.040000000000001,
        "step_limit": 400,
        "steps": 201,
        "success": true,
        "valid": true
      },
      "steps": 201,
      "simulation_time_s": 8.040000000000001,
      "wall_time_s": 775.228552,
      "timing": {
        "simulation_time_s": 8.040000000000001,
        "wall_time_s": 775.228552,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Click the top button of the alarm-clock with beige screen",
      "native_instruction": "Click the top button of the alarm-clock with beige screen",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.956055184801782,
        "cache_read_input_tokens": 857984,
        "cache_reported_input_tokens": 897421,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 897421,
        "cached_input_tokens": 857984,
        "cost_usd": null,
        "fresh_input_tokens": 39437,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 897421,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 857984,
        "known_fresh_input_tokens": 39437,
        "known_input_tokens": 897421,
        "known_output_tokens": 4622,
        "known_reasoning_output_tokens": 567,
        "output_tokens": 4622,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 567,
        "reasoning_reported_output_tokens": 4622,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 902043,
        "uncached_input_tokens": 39437,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11816,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11856,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11989,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12039,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12328,
            "output_tokens": 252,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 17564,
            "output_tokens": 136,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 19769,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 21613,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 22587,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22913,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23103,
            "output_tokens": 54,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23526,
            "output_tokens": 570,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24117,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24318,
            "output_tokens": 54,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24404,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 25392,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25527,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 26665,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 28697,
            "output_tokens": 92,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 30765,
            "output_tokens": 72,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 32810,
            "output_tokens": 81,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33101,
            "output_tokens": 258,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33813,
            "output_tokens": 163,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 35135,
            "output_tokens": 1036,
            "reasoning_output_tokens": 211
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36192,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36909,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 36996,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37578,
            "output_tokens": 113,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38174,
            "output_tokens": 174,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38494,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38587,
            "output_tokens": 72,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39059,
            "output_tokens": 503,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39585,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 42,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 17,
          "bash": 1,
          "exec": 12,
          "devices": 2,
          "device": 1,
          "write": 3,
          "grep": 1,
          "edit": 2,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 201 rosbridge control frames; 35 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-control.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/workspace/.kinex/memos/robotwin-control.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vla_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-14-20261113/workspace/.kinex/tools/vla_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 78,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 35,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 82,
        "published_frames": 82,
        "decode_ok": true,
        "duration_s": 2.05,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "7c6e4595db8fce3c6c5c81157024565de4c49b36aa4082426700921877e6e15c",
        "source_sha256": {
          "third_person": "18f0cd26c8ebbacce015199527bddd85480136d7e9029057889a32d98acd5754",
          "left_wrist": "19deea9f50741dc147073c5a1d7b1e337fb40f17440a3e945ca68374a96d5bb9",
          "right_wrist": "0a9c73f917be9f484181569df2fa7db6e396c0a9d7c81c899025c3b8cb936c36"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "1fba5825b046",
          "task": "robotwin/click-alarmclock-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 82,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 82,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8040000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 82,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 82,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8040000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 82,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 82,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8040000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-15-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/15",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "15",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 6.5600000000000005,
        "step_limit": 400,
        "steps": 164,
        "success": false,
        "valid": true
      },
      "steps": 164,
      "simulation_time_s": 6.5600000000000005,
      "wall_time_s": 755.118069,
      "timing": {
        "simulation_time_s": 6.5600000000000005,
        "wall_time_s": 755.118069,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to press the medium-sized bell's top center.",
      "native_instruction": "Use the left arm to press the medium-sized bell's top center.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9374792715373333,
        "cache_read_input_tokens": 1017600,
        "cache_reported_input_tokens": 1085464,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1085464,
        "cached_input_tokens": 1017600,
        "cost_usd": null,
        "fresh_input_tokens": 67864,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1085464,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1017600,
        "known_fresh_input_tokens": 67864,
        "known_input_tokens": 1085464,
        "known_output_tokens": 6978,
        "known_reasoning_output_tokens": 1690,
        "output_tokens": 6978,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1690,
        "reasoning_reported_output_tokens": 6978,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 37,
          "cache_write_input_tokens": 37,
          "cache_write_reported_input_tokens": 37,
          "cached_input_tokens": 37,
          "fresh_input_tokens": 37,
          "input_tokens": 37,
          "output_tokens": 37,
          "reasoning_output_tokens": 37,
          "reasoning_reported_output_tokens": 37
        },
        "request_attempts": 37,
        "response_count": 37,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1092442,
        "uncached_input_tokens": 67864,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11988,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 16089,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 16333,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 19936,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 21198,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21768,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 21932,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 22122,
            "output_tokens": 197,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 24053,
            "output_tokens": 312,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24900,
            "output_tokens": 321,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26524,
            "output_tokens": 152,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27015,
            "output_tokens": 257,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 28167,
            "output_tokens": 197,
            "reasoning_output_tokens": 122
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 29543,
            "output_tokens": 430,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30581,
            "output_tokens": 489,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31709,
            "output_tokens": 369,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32585,
            "output_tokens": 916,
            "reasoning_output_tokens": 148
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33524,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 33645,
            "output_tokens": 48,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33715,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34202,
            "output_tokens": 119,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34335,
            "output_tokens": 141,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34505,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34959,
            "output_tokens": 622,
            "reasoning_output_tokens": 375
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35697,
            "output_tokens": 90,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 36169,
            "output_tokens": 227,
            "reasoning_output_tokens": 146
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36772,
            "output_tokens": 71,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36899,
            "output_tokens": 184,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37329,
            "output_tokens": 238,
            "reasoning_output_tokens": 202
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 38338,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38379,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38572,
            "output_tokens": 191,
            "reasoning_output_tokens": 110
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39148,
            "output_tokens": 144,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39314,
            "output_tokens": 511,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 39849,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 44,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 14,
          "find": 1,
          "bash": 1,
          "exec": 15,
          "devices": 2,
          "device": 4,
          "write": 2,
          "device_stop": 3,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 164 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-bell.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/workspace/.kinex/memos/robotwin-bell.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/left_bell_motion.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-15-20261113/workspace/.kinex/tools/left_bell_motion.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 84,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 67,
        "published_frames": 67,
        "decode_ok": true,
        "duration_s": 1.675,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c86cfaefdd1f45f307fcc5bb7854babc34300a6800e5386e0e6f7cde74c972f8",
        "source_sha256": {
          "third_person": "153bc00039b24e3e5866c54b138bc4a148abf1cb8ce762888a1c062a0e1f412a",
          "left_wrist": "094bedd28633eadd4488259a06999eadd37c3820215938d03bc1c3ef9bb63254",
          "right_wrist": "bdcf5fe80c6edcefc2a3d7053ae2030358e08ac541de2f4b497b998a46c5b5c6"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "2a5e2de47c71",
          "task": "robotwin/click-bell-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 67,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 67,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 6560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 67,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 67,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 6560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 67,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 67,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 6560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-16-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/16",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "16",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 22.28,
        "step_limit": 600,
        "steps": 557,
        "success": false,
        "valid": true
      },
      "steps": 557,
      "simulation_time_s": 22.28,
      "wall_time_s": 2436.000977,
      "timing": {
        "simulation_time_s": 22.28,
        "wall_time_s": 2436.000977,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the compact tabletop trashbin, pour balls into the bin.",
      "native_instruction": "Pick up the compact tabletop trashbin, pour balls into the bin.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9812236564201158,
        "cache_read_input_tokens": 2889216,
        "cache_reported_input_tokens": 2944503,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2944503,
        "cached_input_tokens": 2889216,
        "cost_usd": null,
        "fresh_input_tokens": 55287,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2944503,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2889216,
        "known_fresh_input_tokens": 55287,
        "known_input_tokens": 2944503,
        "known_output_tokens": 13229,
        "known_reasoning_output_tokens": 4752,
        "output_tokens": 13229,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 4752,
        "reasoning_reported_output_tokens": 13229,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 75,
          "cache_write_input_tokens": 75,
          "cache_write_reported_input_tokens": 75,
          "cached_input_tokens": 75,
          "fresh_input_tokens": 75,
          "input_tokens": 75,
          "output_tokens": 75,
          "reasoning_output_tokens": 75,
          "reasoning_reported_output_tokens": 75
        },
        "request_attempts": 75,
        "response_count": 75,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2957732,
        "uncached_input_tokens": 55287,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11988,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 14194,
            "output_tokens": 174,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14208,
            "input_tokens": 17021,
            "output_tokens": 236,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 21164,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21734,
            "output_tokens": 116,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22964,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23290,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23481,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23889,
            "output_tokens": 857,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24769,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 24974,
            "output_tokens": 110,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25113,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 26002,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26092,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26510,
            "output_tokens": 69,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 28745,
            "output_tokens": 631,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29405,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30392,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31375,
            "output_tokens": 104,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31541,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31793,
            "output_tokens": 276,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32604,
            "output_tokens": 196,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 35136,
            "output_tokens": 1126,
            "reasoning_output_tokens": 285
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36285,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36402,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 37120,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37226,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37654,
            "output_tokens": 101,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38098,
            "output_tokens": 227,
            "reasoning_output_tokens": 149
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38668,
            "output_tokens": 51,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 38770,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 38909,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 38957,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39039,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39307,
            "output_tokens": 111,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39759,
            "output_tokens": 387,
            "reasoning_output_tokens": 297
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40851,
            "output_tokens": 108,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41298,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41491,
            "output_tokens": 317,
            "reasoning_output_tokens": 100
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42250,
            "output_tokens": 446,
            "reasoning_output_tokens": 318
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42785,
            "output_tokens": 189,
            "reasoning_output_tokens": 106
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43677,
            "output_tokens": 77,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44502,
            "output_tokens": 111,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44642,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 45504,
            "output_tokens": 150,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45714,
            "output_tokens": 42,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 45818,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 46221,
            "output_tokens": 55,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46380,
            "output_tokens": 312,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 47098,
            "output_tokens": 300,
            "reasoning_output_tokens": 217
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47741,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 47933,
            "output_tokens": 376,
            "reasoning_output_tokens": 261
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 49207,
            "output_tokens": 436,
            "reasoning_output_tokens": 253
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 49735,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49664,
            "input_tokens": 50162,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50048,
            "input_tokens": 50356,
            "output_tokens": 1393,
            "reasoning_output_tokens": 1034
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51584,
            "input_tokens": 52180,
            "output_tokens": 242,
            "reasoning_output_tokens": 102
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52224,
            "input_tokens": 52761,
            "output_tokens": 49,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52608,
            "input_tokens": 52861,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52736,
            "input_tokens": 52999,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52864,
            "input_tokens": 53047,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52864,
            "input_tokens": 53129,
            "output_tokens": 104,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53120,
            "input_tokens": 53575,
            "output_tokens": 208,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53376,
            "input_tokens": 54261,
            "output_tokens": 251,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54400,
            "input_tokens": 54850,
            "output_tokens": 49,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54656,
            "input_tokens": 54950,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54784,
            "input_tokens": 54998,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54784,
            "input_tokens": 55080,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54912,
            "input_tokens": 55350,
            "output_tokens": 106,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55424,
            "input_tokens": 55796,
            "output_tokens": 271,
            "reasoning_output_tokens": 187
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55936,
            "input_tokens": 56407,
            "output_tokens": 620,
            "reasoning_output_tokens": 585
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56960,
            "input_tokens": 57089,
            "output_tokens": 649,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57600,
            "input_tokens": 57765,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 87,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 19,
          "find": 1,
          "exec": 31,
          "devices": 5,
          "device": 13,
          "write": 3,
          "edit": 4,
          "device_stop": 4,
          "relative_pose": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 557 rosbridge control frames; 26 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_segment.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-16-20261113/workspace/.kinex/tools/policy_segment.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 165,
        "observed_images": 30
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 26,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 224,
        "published_frames": 224,
        "decode_ok": true,
        "duration_s": 5.6,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "da396f14ec07f32277ed2e3d4e17e881b9282289427ea524a3974f513682f212",
        "source_sha256": {
          "third_person": "f4f43cd1ffd75166bd32b25fdfc1786a17faba4790569d8370e0d6e171686601",
          "left_wrist": "9fef8b11cd3a4df3200738682506550267cd7c25a138b6823e1f8eaaea555683",
          "right_wrist": "8ea2f08da6a24e8b99334e546dbe05fca1af2425fbef5521d56f774774df1b1d"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "8945124ed33a",
          "task": "robotwin/dump-bin-bigbin-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 224,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 224,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 22280000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 224,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 224,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 22280000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 224,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 224,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 22280000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-17-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/17",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "17",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.64,
        "step_limit": 400,
        "steps": 241,
        "success": true,
        "valid": true
      },
      "steps": 241,
      "simulation_time_s": 9.64,
      "wall_time_s": 1011.938805,
      "timing": {
        "simulation_time_s": 9.64,
        "wall_time_s": 1011.938805,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use both arms to grab the light brown wooden roller.",
      "native_instruction": "Use both arms to grab the light brown wooden roller.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9594917029977357,
        "cache_read_input_tokens": 567808,
        "cache_reported_input_tokens": 591780,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 591780,
        "cached_input_tokens": 567808,
        "cost_usd": null,
        "fresh_input_tokens": 23972,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 591780,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 567808,
        "known_fresh_input_tokens": 23972,
        "known_input_tokens": 591780,
        "known_output_tokens": 2833,
        "known_reasoning_output_tokens": 145,
        "output_tokens": 2833,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 145,
        "reasoning_reported_output_tokens": 2833,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 24,
          "cache_write_input_tokens": 24,
          "cache_write_reported_input_tokens": 24,
          "cached_input_tokens": 24,
          "fresh_input_tokens": 24,
          "input_tokens": 24,
          "output_tokens": 24,
          "reasoning_output_tokens": 24,
          "reasoning_reported_output_tokens": 24
        },
        "request_attempts": 24,
        "response_count": 24,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 594613,
        "uncached_input_tokens": 23972,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11811,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11851,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11984,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14441,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 18044,
            "output_tokens": 208,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 21919,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 23133,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23459,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23649,
            "output_tokens": 121,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 25114,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25522,
            "output_tokens": 641,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26184,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26373,
            "output_tokens": 103,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26503,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 27318,
            "output_tokens": 43,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 28120,
            "output_tokens": 43,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28949,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29772,
            "output_tokens": 63,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30609,
            "output_tokens": 43,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30843,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30932,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31271,
            "output_tokens": 535,
            "reasoning_output_tokens": 96
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31833,
            "output_tokens": 292,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32146,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 34,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 13,
          "find": 1,
          "exec": 9,
          "devices": 2,
          "device": 1,
          "write": 2,
          "edit": 2,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 241 rosbridge control frames; 61 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/roller.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/workspace/.kinex/memos/roller.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/roller_policy.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-17-20261113/workspace/.kinex/tools/roller_policy.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 61,
        "observed_images": 17
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 61,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 98,
        "published_frames": 98,
        "decode_ok": true,
        "duration_s": 2.45,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "e98f8599d077d610d1629286479d8fca94a3533f10e7c9141c537bec74fbccd3",
        "source_sha256": {
          "third_person": "bebfb73e14db629c1a2950beb5f321c0451b6eb72d9c6829c71c792349135e7f",
          "left_wrist": "12b4255fdc7d6ee5d03c96df98068a225cb6f78a0b7236146a4f26dea98ff0e3",
          "right_wrist": "0b4c4e37b97bafca1730a44c2cd0a8571309e1d72b4f1cac494c31f41326e9bd"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "8c93ce506877",
          "task": "robotwin/grab-roller-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 98,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 98,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 98,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-18-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/18",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "18",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 20.400000000000002,
        "step_limit": 600,
        "steps": 510,
        "success": true,
        "valid": true
      },
      "steps": 510,
      "simulation_time_s": 20.400000000000002,
      "wall_time_s": 2062.476765,
      "timing": {
        "simulation_time_s": 20.400000000000002,
        "wall_time_s": 2062.476765,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the microphone used for audio input and pass it to the other arm.",
      "native_instruction": "Grab the microphone used for audio input and pass it to the other arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9776231210361415,
        "cache_read_input_tokens": 1480576,
        "cache_reported_input_tokens": 1514465,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1514465,
        "cached_input_tokens": 1480576,
        "cost_usd": null,
        "fresh_input_tokens": 33889,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1514465,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1480576,
        "known_fresh_input_tokens": 33889,
        "known_input_tokens": 1514465,
        "known_output_tokens": 8790,
        "known_reasoning_output_tokens": 1681,
        "output_tokens": 8790,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1681,
        "reasoning_reported_output_tokens": 8790,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 50,
          "cache_write_input_tokens": 50,
          "cache_write_reported_input_tokens": 50,
          "cached_input_tokens": 50,
          "fresh_input_tokens": 50,
          "input_tokens": 50,
          "output_tokens": 50,
          "reasoning_output_tokens": 50,
          "reasoning_reported_output_tokens": 50
        },
        "request_attempts": 50,
        "response_count": 50,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1523255,
        "uncached_input_tokens": 33889,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11817,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11857,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11990,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12330,
            "output_tokens": 252,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 17566,
            "output_tokens": 185,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 20730,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21056,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21247,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21655,
            "output_tokens": 665,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22341,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23215,
            "output_tokens": 41,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23318,
            "output_tokens": 245,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23590,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 23796,
            "output_tokens": 129,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 23952,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 25997,
            "output_tokens": 160,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26184,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 27279,
            "output_tokens": 82,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27408,
            "output_tokens": 72,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28414,
            "output_tokens": 81,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28705,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29434,
            "output_tokens": 75,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29975,
            "output_tokens": 1105,
            "reasoning_output_tokens": 231
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31101,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31218,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31701,
            "output_tokens": 107,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32213,
            "output_tokens": 77,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32341,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32479,
            "output_tokens": 193,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32893,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33379,
            "output_tokens": 277,
            "reasoning_output_tokens": 199
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34062,
            "output_tokens": 251,
            "reasoning_output_tokens": 170
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34800,
            "output_tokens": 483,
            "reasoning_output_tokens": 154
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35782,
            "output_tokens": 317,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36432,
            "output_tokens": 245,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37021,
            "output_tokens": 229,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37535,
            "output_tokens": 46,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37751,
            "output_tokens": 290,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38458,
            "output_tokens": 224,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39093,
            "output_tokens": 232,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39627,
            "output_tokens": 215,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40077,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40559,
            "output_tokens": 62,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40683,
            "output_tokens": 192,
            "reasoning_output_tokens": 139
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41288,
            "output_tokens": 128,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41560,
            "output_tokens": 55,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41677,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41870,
            "output_tokens": 258,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42155,
            "output_tokens": 674,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 42854,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 59,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 16,
          "bash": 1,
          "exec": 24,
          "devices": 2,
          "device": 4,
          "write": 3,
          "device_stop": 3,
          "edit": 4,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 510 rosbridge control frames; 45 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/handover-microphone.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/workspace/.kinex/memos/handover-microphone.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_segment.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/workspace/.kinex/tools/cartesian_segment.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_segment.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-18-20261113/workspace/.kinex/tools/policy_segment.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 112,
        "observed_images": 22
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 45,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 205,
        "published_frames": 205,
        "decode_ok": true,
        "duration_s": 5.125,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "815812af60d821fb48733575b1337de4606e33331c29e674583ea872509cd0eb",
        "source_sha256": {
          "third_person": "6791163c5e921d7d5afe9b8358b5ea93d60119ec2cdbd9187113776841993945",
          "left_wrist": "e946d04acfa4d3443881cf181d8e869ccb5972867f4b148bebfc3a647469917f",
          "right_wrist": "8c1bd8a69f7dc93d8f5f7d68cfef416422ca83e97455d578115354ccbc990b1f"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "c19df546d136",
          "task": "robotwin/handover-mic-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 206,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 205,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 20400000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 206,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 205,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 20400000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 206,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 205,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 20400000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-19-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/19",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "19",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.48,
        "step_limit": 400,
        "steps": 312,
        "success": true,
        "valid": true
      },
      "steps": 312,
      "simulation_time_s": 12.48,
      "wall_time_s": 1395.896361,
      "timing": {
        "simulation_time_s": 12.48,
        "wall_time_s": 1395.896361,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the medium sauce can and place it near the smooth gray pot with matte finish",
      "native_instruction": "Grab the medium sauce can and place it near the smooth gray pot with matte finish",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9722742387149335,
        "cache_read_input_tokens": 1419008,
        "cache_reported_input_tokens": 1459473,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1459473,
        "cached_input_tokens": 1419008,
        "cost_usd": null,
        "fresh_input_tokens": 40465,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1459473,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1419008,
        "known_fresh_input_tokens": 40465,
        "known_input_tokens": 1459473,
        "known_output_tokens": 6349,
        "known_reasoning_output_tokens": 1041,
        "output_tokens": 6349,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1041,
        "reasoning_reported_output_tokens": 6349,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 43,
          "cache_write_input_tokens": 43,
          "cache_write_reported_input_tokens": 43,
          "cached_input_tokens": 43,
          "fresh_input_tokens": 43,
          "input_tokens": 43,
          "output_tokens": 43,
          "reasoning_output_tokens": 43,
          "reasoning_reported_output_tokens": 43
        },
        "request_attempts": 43,
        "response_count": 43,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1465822,
        "uncached_input_tokens": 40465,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11859,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11992,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14893,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 15057,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15104,
            "input_tokens": 18660,
            "output_tokens": 239,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 25300,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 26670,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26996,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27187,
            "output_tokens": 62,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27618,
            "output_tokens": 644,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28284,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 29162,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29251,
            "output_tokens": 277,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29556,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29763,
            "output_tokens": 128,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29919,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 31436,
            "output_tokens": 81,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 35014,
            "output_tokens": 92,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35134,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35815,
            "output_tokens": 160,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36753,
            "output_tokens": 287,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37433,
            "output_tokens": 99,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37738,
            "output_tokens": 287,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38592,
            "output_tokens": 941,
            "reasoning_output_tokens": 297
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39554,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39669,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40438,
            "output_tokens": 117,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 41239,
            "output_tokens": 122,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41429,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41641,
            "output_tokens": 151,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41860,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42360,
            "output_tokens": 148,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42925,
            "output_tokens": 65,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43353,
            "output_tokens": 117,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43876,
            "output_tokens": 191,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44483,
            "output_tokens": 162,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 45055,
            "output_tokens": 78,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45495,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45682,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 45773,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 46040,
            "output_tokens": 637,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 46700,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 54,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 16,
          "find": 1,
          "bash": 1,
          "exec": 23,
          "devices": 2,
          "device": 1,
          "write": 3,
          "device_stop": 2,
          "edit": 3,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 312 rosbridge control frames; 35 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-control.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/workspace/.kinex/memos/robotwin-control.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vla_rollout.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-19-20261113/workspace/.kinex/tools/vla_rollout.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 100,
        "observed_images": 18
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 35,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 126,
        "published_frames": 126,
        "decode_ok": true,
        "duration_s": 3.15,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "cc71ab75163bad442587d610615dae013ef1b8a2ce4930d6cc427dc064e76cd0",
        "source_sha256": {
          "third_person": "8abc579222dab82f3b3898b1d9a3d0afadb9b9f8b949c806caefedd62d40b9bc",
          "left_wrist": "c7097c65b94d32d59fc6ad276215090c1ccd2326f649376acaff3676086853f7",
          "right_wrist": "55756d87553a83f17f1495b862a64a77187735d8f2ddd6e5d171faddf3d2b897"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "1ee0a0802f58",
          "task": "robotwin/move-can-pot-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 126,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 126,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12480000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 126,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 126,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12480000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 126,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 126,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12480000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-20-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/20",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "20",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 13.48,
        "step_limit": 400,
        "steps": 337,
        "success": true,
        "valid": true
      },
      "steps": 337,
      "simulation_time_s": 13.48,
      "wall_time_s": 1258.173439,
      "timing": {
        "simulation_time_s": 13.48,
        "wall_time_s": 1258.173439,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the medium-sized bottle and put it onto the pad",
      "native_instruction": "Pick the medium-sized bottle and put it onto the pad",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9465677778091676,
        "cache_read_input_tokens": 1139200,
        "cache_reported_input_tokens": 1203506,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1203506,
        "cached_input_tokens": 1139200,
        "cost_usd": null,
        "fresh_input_tokens": 64306,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1203506,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1139200,
        "known_fresh_input_tokens": 64306,
        "known_input_tokens": 1203506,
        "known_output_tokens": 6151,
        "known_reasoning_output_tokens": 747,
        "output_tokens": 6151,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 747,
        "reasoning_reported_output_tokens": 6151,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 41,
          "cache_write_input_tokens": 41,
          "cache_write_reported_input_tokens": 41,
          "cached_input_tokens": 41,
          "fresh_input_tokens": 41,
          "input_tokens": 41,
          "output_tokens": 41,
          "reasoning_output_tokens": 41,
          "reasoning_reported_output_tokens": 41
        },
        "request_attempts": 41,
        "response_count": 41,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1209657,
        "uncached_input_tokens": 64306,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11991,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12039,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12328,
            "output_tokens": 252,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 17564,
            "output_tokens": 190,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 21307,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21633,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 23049,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23240,
            "output_tokens": 57,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23666,
            "output_tokens": 532,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24220,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24424,
            "output_tokens": 41,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24527,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24653,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25490,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25579,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25996,
            "output_tokens": 59,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 28064,
            "output_tokens": 690,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28776,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29867,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 31127,
            "output_tokens": 75,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 32423,
            "output_tokens": 155,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33945,
            "output_tokens": 274,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34573,
            "output_tokens": 429,
            "reasoning_output_tokens": 209
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35359,
            "output_tokens": 350,
            "reasoning_output_tokens": 102
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35974,
            "output_tokens": 728,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36723,
            "output_tokens": 47,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 36832,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37266,
            "output_tokens": 108,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37726,
            "output_tokens": 64,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38060,
            "output_tokens": 117,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38522,
            "output_tokens": 187,
            "reasoning_output_tokens": 96
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39395,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39454,
            "output_tokens": 112,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39916,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40354,
            "output_tokens": 72,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40568,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 40662,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40971,
            "output_tokens": 572,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41567,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 52,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 17,
          "bash": 1,
          "exec": 18,
          "devices": 3,
          "device": 1,
          "write": 4,
          "device_stop": 5,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 337 rosbridge control frames; 42 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin_bottle.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/workspace/.kinex/memos/robotwin_bottle.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vla_rollout.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-20-20261113/workspace/.kinex/tools/vla_rollout.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 96,
        "observed_images": 18
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 42,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 136,
        "published_frames": 136,
        "decode_ok": true,
        "duration_s": 3.4,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c54e7a6b1b7ea1504699f180fb3d4ac31c71bc85a4bc4d0e7c433ad0b4dc86db",
        "source_sha256": {
          "third_person": "82c7fdcae92c329057a12888f3e02bdd3c48078fbe63557ce41fade1d108d052",
          "left_wrist": "c4dda31c3315ee1fff1b83831c7978048bd7fd437fadb32a5d6babf6d77bdfa0",
          "right_wrist": "f87a88ae43771e7606f4e11fa83265314c5226bf19916d437efa5c409c02dc42"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ae55459d58d5",
          "task": "robotwin/move-pillbottle-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 136,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 136,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13480000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 136,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 136,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13480000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 136,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 136,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13480000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-21-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/21",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "21",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 13.68,
        "step_limit": 400,
        "steps": 342,
        "success": false,
        "valid": true
      },
      "steps": 342,
      "simulation_time_s": 13.68,
      "wall_time_s": 1251.544861,
      "timing": {
        "simulation_time_s": 13.68,
        "wall_time_s": 1251.544861,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the small packet of playingcards and move it outward.",
      "native_instruction": "Pick up the small packet of playingcards and move it outward.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9254150105213935,
        "cache_read_input_tokens": 1013248,
        "cache_reported_input_tokens": 1094912,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1094912,
        "cached_input_tokens": 1013248,
        "cost_usd": null,
        "fresh_input_tokens": 81664,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1094912,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1013248,
        "known_fresh_input_tokens": 81664,
        "known_input_tokens": 1094912,
        "known_output_tokens": 5487,
        "known_reasoning_output_tokens": 685,
        "output_tokens": 5487,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 685,
        "reasoning_reported_output_tokens": 5487,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 37,
          "cache_write_input_tokens": 37,
          "cache_write_reported_input_tokens": 37,
          "cached_input_tokens": 37,
          "fresh_input_tokens": 37,
          "input_tokens": 37,
          "output_tokens": 37,
          "reasoning_output_tokens": 37,
          "reasoning_reported_output_tokens": 37
        },
        "request_attempts": 37,
        "response_count": 37,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1100399,
        "uncached_input_tokens": 81664,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11817,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11950,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11998,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12287,
            "output_tokens": 252,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 17523,
            "output_tokens": 194,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 21270,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21596,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 23013,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23205,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23613,
            "output_tokens": 604,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24238,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25075,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25166,
            "output_tokens": 352,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25545,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25750,
            "output_tokens": 106,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25883,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 27947,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 29950,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 31956,
            "output_tokens": 138,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32575,
            "output_tokens": 329,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33832,
            "output_tokens": 793,
            "reasoning_output_tokens": 162
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34646,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34802,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34959,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35050,
            "output_tokens": 116,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 36712,
            "output_tokens": 170,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 36909,
            "output_tokens": 101,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37440,
            "output_tokens": 120,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37993,
            "output_tokens": 64,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38489,
            "output_tokens": 70,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38581,
            "output_tokens": 143,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 39156,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39636,
            "output_tokens": 128,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 40198,
            "output_tokens": 199,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40832,
            "output_tokens": 120,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41383,
            "output_tokens": 529,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 41937,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 50,
        "model_tool_calls_by_name": {
          "read": 16,
          "ls": 1,
          "bash": 1,
          "exec": 18,
          "devices": 2,
          "device": 2,
          "write": 3,
          "device_stop": 3,
          "edit": 3,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 342 rosbridge control frames; 30 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/aloha_manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/workspace/.kinex/memos/aloha_manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-21-20261113/workspace/.kinex/tools/policy_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 90,
        "observed_images": 20
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 30,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 138,
        "published_frames": 138,
        "decode_ok": true,
        "duration_s": 3.45,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "b7d797a2fafc329717dbfdf5d48e122a855dfec835163e3c48ce0c3d4daf49f9",
        "source_sha256": {
          "third_person": "3c64b4df366f501c0a8fa89a72189cfaec14c8bbd1f8668c4d2bf1c16066899b",
          "left_wrist": "f3a8d6fe375fb1cdf04fc42e6eb2273daa8193329373f6de70473bc0c9234116",
          "right_wrist": "05c83f43bd60b4d6b09bbc263e15e2a178ab54b86e13d3e18a216ee50e58bbb8"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "a4837c95b036",
          "task": "robotwin/move-playingcard-away-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 138,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13680000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 138,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13680000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 138,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13680000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-22-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/22",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "22",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.72,
        "step_limit": 400,
        "steps": 318,
        "success": true,
        "valid": true
      },
      "steps": 318,
      "simulation_time_s": 12.72,
      "wall_time_s": 1072.231804,
      "timing": {
        "simulation_time_s": 12.72,
        "wall_time_s": 1072.231804,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the red plastic stapler and set it on Red mat",
      "native_instruction": "Grab the red plastic stapler and set it on Red mat",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9731137031553352,
        "cache_read_input_tokens": 1131776,
        "cache_reported_input_tokens": 1163046,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1163046,
        "cached_input_tokens": 1131776,
        "cost_usd": null,
        "fresh_input_tokens": 31270,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1163046,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1131776,
        "known_fresh_input_tokens": 31270,
        "known_input_tokens": 1163046,
        "known_output_tokens": 6083,
        "known_reasoning_output_tokens": 807,
        "output_tokens": 6083,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 807,
        "reasoning_reported_output_tokens": 6083,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 40,
          "cache_write_input_tokens": 40,
          "cache_write_reported_input_tokens": 40,
          "cached_input_tokens": 40,
          "fresh_input_tokens": 40,
          "input_tokens": 40,
          "output_tokens": 40,
          "reasoning_output_tokens": 40,
          "reasoning_reported_output_tokens": 40
        },
        "request_attempts": 40,
        "response_count": 40,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1169129,
        "uncached_input_tokens": 31270,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11991,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12331,
            "output_tokens": 205,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 15946,
            "output_tokens": 259,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 20939,
            "output_tokens": 158,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21315,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21723,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 21915,
            "output_tokens": 62,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22298,
            "output_tokens": 717,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23039,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23920,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24012,
            "output_tokens": 242,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24284,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24492,
            "output_tokens": 51,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 26485,
            "output_tokens": 121,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26636,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27794,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28943,
            "output_tokens": 75,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 30107,
            "output_tokens": 93,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30262,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30514,
            "output_tokens": 277,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31199,
            "output_tokens": 175,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 33710,
            "output_tokens": 1125,
            "reasoning_output_tokens": 241
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 34858,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35019,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35532,
            "output_tokens": 131,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36061,
            "output_tokens": 107,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36570,
            "output_tokens": 55,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36647,
            "output_tokens": 80,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37128,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37601,
            "output_tokens": 294,
            "reasoning_output_tokens": 165
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38295,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38788,
            "output_tokens": 113,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39303,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39496,
            "output_tokens": 45,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39603,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39870,
            "output_tokens": 172,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40072,
            "output_tokens": 573,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40672,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 51,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 18,
          "bash": 1,
          "exec": 17,
          "devices": 3,
          "device": 1,
          "write": 3,
          "device_stop": 4,
          "edit": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 318 rosbridge control frames; 30 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_rollout.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-22-20261113/workspace/.kinex/tools/policy_rollout.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 94,
        "observed_images": 21
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 30,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 128,
        "published_frames": 128,
        "decode_ok": true,
        "duration_s": 3.2,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f877db59491885bdb89b3b4f69dde61ff8675d30f54a573fe51a225b14c4cc0c",
        "source_sha256": {
          "third_person": "311ac4c69392aa6c9900a605bc8655684e92a497c4bbfad4f28f98d0ab3703ad",
          "left_wrist": "4a28be81bbe9574e511821a84e87be1152dc4dcb0117f06987bed3db2dc2437f",
          "right_wrist": "c60d0d12cfcedae27ffc86d204a9042e414d3419830b4da62653f4b9bdff8d19"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "2c9416865d1b",
          "task": "robotwin/move-stapler-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 129,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 128,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12720000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 129,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 128,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12720000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 129,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 128,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12720000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-23-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/23",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "23",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 26.6,
        "step_limit": 700,
        "steps": 665,
        "success": false,
        "valid": true
      },
      "steps": 665,
      "simulation_time_s": 26.6,
      "wall_time_s": 2513.676409,
      "timing": {
        "simulation_time_s": 26.6,
        "wall_time_s": 2513.676409,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift the black laptop using the right arm.",
      "native_instruction": "Lift the black laptop using the right arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9745476508613936,
        "cache_read_input_tokens": 1710336,
        "cache_reported_input_tokens": 1755005,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1755005,
        "cached_input_tokens": 1710336,
        "cost_usd": null,
        "fresh_input_tokens": 44669,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1755005,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1710336,
        "known_fresh_input_tokens": 44669,
        "known_input_tokens": 1755005,
        "known_output_tokens": 9811,
        "known_reasoning_output_tokens": 2835,
        "output_tokens": 9811,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2835,
        "reasoning_reported_output_tokens": 9811,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 55,
          "cache_write_input_tokens": 55,
          "cache_write_reported_input_tokens": 55,
          "cached_input_tokens": 55,
          "fresh_input_tokens": 55,
          "input_tokens": 55,
          "output_tokens": 55,
          "reasoning_output_tokens": 55,
          "reasoning_reported_output_tokens": 55
        },
        "request_attempts": 55,
        "response_count": 55,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1764816,
        "uncached_input_tokens": 44669,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11809,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11849,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11982,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12032,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13636,
            "output_tokens": 191,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13696,
            "input_tokens": 17244,
            "output_tokens": 228,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 21132,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 22106,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22432,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22623,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 23031,
            "output_tokens": 685,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 23739,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 23938,
            "output_tokens": 59,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24029,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 25048,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25202,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26032,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26119,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26534,
            "output_tokens": 382,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26943,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 28131,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 29321,
            "output_tokens": 88,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29469,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29559,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29866,
            "output_tokens": 77,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 31098,
            "output_tokens": 109,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31417,
            "output_tokens": 277,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32240,
            "output_tokens": 74,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32780,
            "output_tokens": 469,
            "reasoning_output_tokens": 232
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33752,
            "output_tokens": 254,
            "reasoning_output_tokens": 198
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 35319,
            "output_tokens": 577,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35958,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 37177,
            "output_tokens": 915,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38112,
            "output_tokens": 163,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38453,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38648,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38742,
            "output_tokens": 47,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 38826,
            "output_tokens": 84,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39247,
            "output_tokens": 144,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39417,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39805,
            "output_tokens": 193,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40290,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40346,
            "output_tokens": 330,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40910,
            "output_tokens": 463,
            "reasoning_output_tokens": 425
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 42068,
            "output_tokens": 381,
            "reasoning_output_tokens": 146
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42689,
            "output_tokens": 495,
            "reasoning_output_tokens": 371
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 43251,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43678,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 44107,
            "output_tokens": 160,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44563,
            "output_tokens": 174,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 44823,
            "output_tokens": 208,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45326,
            "output_tokens": 296,
            "reasoning_output_tokens": 263
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45684,
            "output_tokens": 235,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 45945,
            "output_tokens": 559,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46528,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 65,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 18,
          "find": 1,
          "exec": 22,
          "devices": 4,
          "device": 2,
          "write": 3,
          "grep": 1,
          "edit": 4,
          "device_stop": 6
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 665 rosbridge control frames; 92 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-laptop.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/workspace/.kinex/memos/robotwin-laptop.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/workspace/.kinex/tools/policy_step.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/right_tcp.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-23-20261113/workspace/.kinex/tools/right_tcp.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 123,
        "observed_images": 23
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 92,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 267,
        "published_frames": 267,
        "decode_ok": true,
        "duration_s": 6.675,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c2225a8cd28d3c9b6dd7959051db820c150cf8076586dbaf488a8262a9d3acfb",
        "source_sha256": {
          "third_person": "cc7413b452eebbcfb5e56b5313f1d2360fb669f7a9f86cc6e884a4ec8583013e",
          "left_wrist": "97db3924f6a1fc01e09ff1066320bdfc894a943900e96f254353f39ebd6d1aa2",
          "right_wrist": "623a5ce20d6d6c18554775d00e4aed0802cd39a65e31bf28a6ae45b6a485960c"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "0df2c4c915f7",
          "task": "robotwin/open-laptop-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 268,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 267,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 26600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 268,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 267,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 26600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 268,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 267,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 26600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-24-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/24",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "24",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 27.72,
        "step_limit": 1500,
        "steps": 693,
        "success": true,
        "valid": true
      },
      "steps": 693,
      "simulation_time_s": 27.72,
      "wall_time_s": 1035.141621,
      "timing": {
        "simulation_time_s": 27.72,
        "wall_time_s": 1035.141621,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to pull the handle of the kitchen microwave with slanted shape.",
      "native_instruction": "Use the left arm to pull the handle of the kitchen microwave with slanted shape.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9556043822214554,
        "cache_read_input_tokens": 753280,
        "cache_reported_input_tokens": 788276,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 788276,
        "cached_input_tokens": 753280,
        "cost_usd": null,
        "fresh_input_tokens": 34996,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 788276,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 753280,
        "known_fresh_input_tokens": 34996,
        "known_input_tokens": 788276,
        "known_output_tokens": 5312,
        "known_reasoning_output_tokens": 878,
        "output_tokens": 5312,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 878,
        "reasoning_reported_output_tokens": 5312,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 31,
          "cache_write_input_tokens": 31,
          "cache_write_reported_input_tokens": 31,
          "cached_input_tokens": 31,
          "fresh_input_tokens": 31,
          "input_tokens": 31,
          "output_tokens": 31,
          "reasoning_output_tokens": 31,
          "reasoning_reported_output_tokens": 31
        },
        "request_attempts": 31,
        "response_count": 31,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 793588,
        "uncached_input_tokens": 34996,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11820,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11953,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12007,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12296,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 15899,
            "output_tokens": 189,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 19731,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 21101,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21427,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21618,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 22026,
            "output_tokens": 590,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22637,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22844,
            "output_tokens": 129,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23000,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23800,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24637,
            "output_tokens": 92,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25470,
            "output_tokens": 60,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25592,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25844,
            "output_tokens": 277,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26806,
            "output_tokens": 176,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 29318,
            "output_tokens": 1029,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30368,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 31043,
            "output_tokens": 104,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31742,
            "output_tokens": 202,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32533,
            "output_tokens": 192,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32946,
            "output_tokens": 203,
            "reasoning_output_tokens": 113
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33745,
            "output_tokens": 417,
            "reasoning_output_tokens": 297
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34755,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34985,
            "output_tokens": 57,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35104,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35296,
            "output_tokens": 614,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 35933,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 41,
        "model_tool_calls_by_name": {
          "read": 17,
          "ls": 1,
          "bash": 1,
          "exec": 13,
          "devices": 1,
          "device": 2,
          "write": 3,
          "edit": 1,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 693 rosbridge control frames; 75 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/open_microwave.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/workspace/.kinex/memos/open_microwave.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_batch.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-24-20261113/workspace/.kinex/tools/policy_batch.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 75,
        "observed_images": 22
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1500,
        "max_control_steps": 1500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 75,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 278,
        "published_frames": 278,
        "decode_ok": true,
        "duration_s": 6.95,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "2a63bf0ef84f3c3057f015ea4850d715c0bbdc6933d4417fdb1c10b9dcedd1a5",
        "source_sha256": {
          "third_person": "f5a5ef015d6c8bda8be5b757a37d805aa96afb31c706f50f4ce7852a07e40854",
          "left_wrist": "446a54016a48557da027e11f08288799515042277d2b48ae8afbfec75d3da590",
          "right_wrist": "7bf0f5065f3ddd55c1bc168fc7b677b6690e2b420f2de33abbbc14c2b92d3b40"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "6bae455bcefa",
          "task": "robotwin/open-microwave-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 279,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 278,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27720000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 279,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 278,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27720000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 279,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 278,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27720000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-25-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/25",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "25",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 2
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.56,
        "step_limit": 400,
        "steps": 389,
        "success": false,
        "valid": true
      },
      "steps": 389,
      "simulation_time_s": 15.56,
      "wall_time_s": 2093.103339,
      "timing": {
        "simulation_time_s": 15.56,
        "wall_time_s": 2093.103339,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift the bottle with attached loop on cap and the plastic bottle with blue label simultaneously with separate arms.",
      "native_instruction": "Lift the bottle with attached loop on cap and the plastic bottle with blue label simultaneously with separate arms.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9792020265286471,
        "cache_read_input_tokens": 2112128,
        "cache_reported_input_tokens": 2156989,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2156989,
        "cached_input_tokens": 2112128,
        "cost_usd": null,
        "fresh_input_tokens": 44861,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2156989,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2112128,
        "known_fresh_input_tokens": 44861,
        "known_input_tokens": 2156989,
        "known_output_tokens": 9590,
        "known_reasoning_output_tokens": 2884,
        "output_tokens": 9590,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2884,
        "reasoning_reported_output_tokens": 9590,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 65,
          "cache_write_input_tokens": 65,
          "cache_write_reported_input_tokens": 65,
          "cached_input_tokens": 65,
          "fresh_input_tokens": 65,
          "input_tokens": 65,
          "output_tokens": 65,
          "reasoning_output_tokens": 65,
          "reasoning_reported_output_tokens": 65
        },
        "request_attempts": 65,
        "response_count": 65,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2166579,
        "uncached_input_tokens": 44861,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11830,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11870,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12003,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12053,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13321,
            "output_tokens": 202,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13440,
            "input_tokens": 15676,
            "output_tokens": 114,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15616,
            "input_tokens": 17902,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18127,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18697,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 20899,
            "output_tokens": 120,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21237,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 22212,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22620,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22809,
            "output_tokens": 563,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23395,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23606,
            "output_tokens": 60,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 23714,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24420,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24547,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25255,
            "output_tokens": 65,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26084,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26174,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26414,
            "output_tokens": 230,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26691,
            "output_tokens": 209,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26929,
            "output_tokens": 65,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27789,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28816,
            "output_tokens": 76,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29640,
            "output_tokens": 93,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30427,
            "output_tokens": 66,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30703,
            "output_tokens": 250,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31401,
            "output_tokens": 1034,
            "reasoning_output_tokens": 263
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32459,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33240,
            "output_tokens": 97,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 34044,
            "output_tokens": 112,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34867,
            "output_tokens": 71,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 34960,
            "output_tokens": 254,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35425,
            "output_tokens": 171,
            "reasoning_output_tokens": 118
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36305,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 37071,
            "output_tokens": 98,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37635,
            "output_tokens": 175,
            "reasoning_output_tokens": 138
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37832,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38289,
            "output_tokens": 129,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 39126,
            "output_tokens": 92,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39931,
            "output_tokens": 72,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40025,
            "output_tokens": 87,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40822,
            "output_tokens": 396,
            "reasoning_output_tokens": 198
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41480,
            "output_tokens": 298,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41808,
            "output_tokens": 199,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42694,
            "output_tokens": 370,
            "reasoning_output_tokens": 277
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43751,
            "output_tokens": 65,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44775,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45629,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 45747,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 46564,
            "output_tokens": 114,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 46708,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 47539,
            "output_tokens": 319,
            "reasoning_output_tokens": 237
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 48569,
            "output_tokens": 229,
            "reasoning_output_tokens": 190
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48768,
            "input_tokens": 48820,
            "output_tokens": 337,
            "reasoning_output_tokens": 145
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49367,
            "output_tokens": 259,
            "reasoning_output_tokens": 218
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 50321,
            "output_tokens": 155,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50432,
            "input_tokens": 51186,
            "output_tokens": 73,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51072,
            "input_tokens": 51965,
            "output_tokens": 195,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51968,
            "input_tokens": 52866,
            "output_tokens": 80,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52864,
            "input_tokens": 53658,
            "output_tokens": 565,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54144,
            "input_tokens": 54250,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 72,
        "model_tool_calls_by_name": {
          "ls": 5,
          "read": 16,
          "find": 1,
          "exec": 32,
          "devices": 2,
          "device": 1,
          "write": 3,
          "grep": 1,
          "edit": 6,
          "device_stop": 5
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 389 rosbridge control frames; 41 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-bottles.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/workspace/.kinex/memos/robotwin-bottles.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_cartesian.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/workspace/.kinex/tools/dual_cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vla_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-25-20261113/workspace/.kinex/tools/vla_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 140,
        "observed_images": 48
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 41,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 157,
        "published_frames": 157,
        "decode_ok": true,
        "duration_s": 3.925,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "acdf5d846a10241873e5cdf17c0dd2e6d72a074aff46bec001da369c18a90bad",
        "source_sha256": {
          "third_person": "ee8b39fbba6375107535e61e9f33a47fd9169bfa72d20d1c902918f7b95b87be",
          "left_wrist": "3aa58fb45199dcf25f5b0d4efb8615a2e125ab87acaa3b491b7e247fa00a6f12",
          "right_wrist": "f962b5885fbecc1c1304094a5ba93104a1e62552a9ed2b47c8d78997af923872"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "e673a2ae52a6",
          "task": "robotwin/pick-diverse-bottles-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 157,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 157,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 157,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 157,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 157,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 157,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-26-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/26",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "26",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 10.72,
        "step_limit": 400,
        "steps": 268,
        "success": false,
        "valid": true
      },
      "steps": 268,
      "simulation_time_s": 10.72,
      "wall_time_s": 1347.616935,
      "timing": {
        "simulation_time_s": 10.72,
        "wall_time_s": 1347.616935,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use one arm for the bottle with red screw cap and the other for the green bottle.",
      "native_instruction": "Use one arm for the bottle with red screw cap and the other for the green bottle.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9715430171207063,
        "cache_read_input_tokens": 902272,
        "cache_reported_input_tokens": 928700,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 928700,
        "cached_input_tokens": 902272,
        "cost_usd": null,
        "fresh_input_tokens": 26428,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 928700,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 902272,
        "known_fresh_input_tokens": 26428,
        "known_input_tokens": 928700,
        "known_output_tokens": 6073,
        "known_reasoning_output_tokens": 783,
        "output_tokens": 6073,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 783,
        "reasoning_reported_output_tokens": 6073,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 36,
          "cache_write_input_tokens": 36,
          "cache_write_reported_input_tokens": 36,
          "cached_input_tokens": 36,
          "fresh_input_tokens": 36,
          "input_tokens": 36,
          "output_tokens": 36,
          "reasoning_output_tokens": 36,
          "reasoning_reported_output_tokens": 36
        },
        "request_attempts": 36,
        "response_count": 36,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 934773,
        "uncached_input_tokens": 26428,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11824,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11955,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12003,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12292,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12443,
            "output_tokens": 226,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 17534,
            "output_tokens": 136,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 19739,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 21160,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21486,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 21676,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 22084,
            "output_tokens": 763,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 22868,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23075,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23165,
            "output_tokens": 114,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23306,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 24184,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24275,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24870,
            "output_tokens": 888,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25779,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26620,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27458,
            "output_tokens": 79,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28316,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 29165,
            "output_tokens": 165,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 30697,
            "output_tokens": 282,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31454,
            "output_tokens": 946,
            "reasoning_output_tokens": 230
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32422,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32853,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33689,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33881,
            "output_tokens": 192,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34207,
            "output_tokens": 118,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34385,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34479,
            "output_tokens": 101,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 35236,
            "output_tokens": 193,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35643,
            "output_tokens": 290,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 35961,
            "output_tokens": 531,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36516,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 45,
        "model_tool_calls_by_name": {
          "read": 16,
          "ls": 1,
          "bash": 1,
          "exec": 13,
          "devices": 3,
          "device": 2,
          "write": 4,
          "device_stop": 3,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 268 rosbridge control frames; 42 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/dual_bottles.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/workspace/.kinex/memos/dual_bottles.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/bottle_policy.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/workspace/.kinex/tools/bottle_policy.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_tcp_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-26-20261113/workspace/.kinex/tools/dual_tcp_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 84,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 42,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 108,
        "published_frames": 108,
        "decode_ok": true,
        "duration_s": 2.7,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "7b8347be052c3bb1d9f88b710e9fc22685dba9e8768fdf28c523f64a552db993",
        "source_sha256": {
          "third_person": "da1b4130147c6a1fa9937c9dc269aa45edd593c48eafcc111dfa368226b50489",
          "left_wrist": "f33e304bbedc0263148f7a5438c1288a4aea9775c64286870cbc9fa47eb85eb2",
          "right_wrist": "60d5a9942872568378e7dbaf27ac7dd22c11550657740f3ce1e733ca1aadc86f"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "116caf1df148",
          "task": "robotwin/pick-dual-bottles-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 109,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 108,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10720000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 109,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 108,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10720000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 109,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 108,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10720000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-27-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/27",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "27",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.24,
        "step_limit": 400,
        "steps": 306,
        "success": true,
        "valid": true
      },
      "steps": 306,
      "simulation_time_s": 12.24,
      "wall_time_s": 1228.171923,
      "timing": {
        "simulation_time_s": 12.24,
        "wall_time_s": 1228.171923,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Position the coffee-box with mug image on the left of the mouse with scroll wheel",
      "native_instruction": "Position the coffee-box with mug image on the left of the mouse with scroll wheel",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9747271125377078,
        "cache_read_input_tokens": 1054336,
        "cache_reported_input_tokens": 1081673,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1081673,
        "cached_input_tokens": 1054336,
        "cost_usd": null,
        "fresh_input_tokens": 27337,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1081673,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1054336,
        "known_fresh_input_tokens": 27337,
        "known_input_tokens": 1081673,
        "known_output_tokens": 5261,
        "known_reasoning_output_tokens": 679,
        "output_tokens": 5261,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 679,
        "reasoning_reported_output_tokens": 5261,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 42,
          "cache_write_input_tokens": 42,
          "cache_write_reported_input_tokens": 42,
          "cached_input_tokens": 42,
          "fresh_input_tokens": 42,
          "input_tokens": 42,
          "output_tokens": 42,
          "reasoning_output_tokens": 42,
          "reasoning_reported_output_tokens": 42
        },
        "request_attempts": 42,
        "response_count": 42,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1086934,
        "uncached_input_tokens": 27337,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11991,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12041,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12330,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12481,
            "output_tokens": 167,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 15939,
            "output_tokens": 198,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 19780,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 20964,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21290,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21480,
            "output_tokens": 62,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 21911,
            "output_tokens": 558,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22490,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 23326,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23415,
            "output_tokens": 62,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23825,
            "output_tokens": 284,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24136,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24342,
            "output_tokens": 53,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25954,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26076,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26483,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26887,
            "output_tokens": 91,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27320,
            "output_tokens": 130,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28353,
            "output_tokens": 56,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28579,
            "output_tokens": 343,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29361,
            "output_tokens": 87,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29914,
            "output_tokens": 776,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30711,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30826,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31215,
            "output_tokens": 143,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31672,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32027,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32386,
            "output_tokens": 118,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32813,
            "output_tokens": 218,
            "reasoning_output_tokens": 139
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33342,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33730,
            "output_tokens": 70,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34111,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34301,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34393,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34839,
            "output_tokens": 555,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35419,
            "output_tokens": 98,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35544,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 50,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 17,
          "bash": 1,
          "exec": 17,
          "devices": 3,
          "device": 2,
          "write": 3,
          "device_stop": 2,
          "edit": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 306 rosbridge control frames; 30 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/workspace/.kinex/tools/cartesian_step.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vla_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-27-20261113/workspace/.kinex/tools/vla_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 95,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 30,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 124,
        "published_frames": 124,
        "decode_ok": true,
        "duration_s": 3.1,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "8493a2de111af79052813f75840893d3e471b572586215c76ef15342da725971",
        "source_sha256": {
          "third_person": "e655d5691eb4a42556d0c9e38266f29894659b88ea858b19129340cf52327531",
          "left_wrist": "6697c01e2796537d12e1a03c269aaa7b80231cf58850276e6dcc3c9cac855af8",
          "right_wrist": "dff49c868b6908065c6c75fb2065acdb4036c7255480a3526b8d0a9f0015a53e"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "506160ea39bb",
          "task": "robotwin/place-a2b-left-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 124,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 124,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12240000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 124,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 124,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12240000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 124,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 124,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12240000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-28-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/28",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "28",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.88,
        "step_limit": 400,
        "steps": 397,
        "success": false,
        "valid": true
      },
      "steps": 397,
      "simulation_time_s": 15.88,
      "wall_time_s": 1605.055426,
      "timing": {
        "simulation_time_s": 15.88,
        "wall_time_s": 1605.055426,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Put the tiny pink car with four wheels to the right of the red, blue, green, yellow rubikscube",
      "native_instruction": "Put the tiny pink car with four wheels to the right of the red, blue, green, yellow rubikscube",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9831117302793769,
        "cache_read_input_tokens": 2303360,
        "cache_reported_input_tokens": 2342928,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2342928,
        "cached_input_tokens": 2303360,
        "cost_usd": null,
        "fresh_input_tokens": 39568,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2342928,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2303360,
        "known_fresh_input_tokens": 39568,
        "known_input_tokens": 2342928,
        "known_output_tokens": 9534,
        "known_reasoning_output_tokens": 2151,
        "output_tokens": 9534,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2151,
        "reasoning_reported_output_tokens": 9534,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 70,
          "cache_write_input_tokens": 70,
          "cache_write_reported_input_tokens": 70,
          "cached_input_tokens": 70,
          "fresh_input_tokens": 70,
          "input_tokens": 70,
          "output_tokens": 70,
          "reasoning_output_tokens": 70,
          "reasoning_reported_output_tokens": 70
        },
        "request_attempts": 70,
        "response_count": 70,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2352462,
        "uncached_input_tokens": 39568,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11829,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11962,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12010,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12299,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12762,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12672,
            "input_tokens": 13333,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13184,
            "input_tokens": 13484,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13312,
            "input_tokens": 14298,
            "output_tokens": 198,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14336,
            "input_tokens": 19156,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 20700,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 21270,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21596,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 22571,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22762,
            "output_tokens": 84,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23215,
            "output_tokens": 605,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 23841,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24724,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24812,
            "output_tokens": 283,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25122,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25336,
            "output_tokens": 112,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25475,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26824,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 29037,
            "output_tokens": 91,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 31261,
            "output_tokens": 61,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31490,
            "output_tokens": 112,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31812,
            "output_tokens": 182,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32248,
            "output_tokens": 260,
            "reasoning_output_tokens": 221
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32974,
            "output_tokens": 641,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33636,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33751,
            "output_tokens": 153,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34222,
            "output_tokens": 46,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34319,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34457,
            "output_tokens": 367,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35077,
            "output_tokens": 90,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35338,
            "output_tokens": 208,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36266,
            "output_tokens": 106,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36697,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37109,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37302,
            "output_tokens": 181,
            "reasoning_output_tokens": 156
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37651,
            "output_tokens": 117,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38087,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38169,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38308,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38356,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38436,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 39244,
            "output_tokens": 107,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39671,
            "output_tokens": 269,
            "reasoning_output_tokens": 176
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40256,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40446,
            "output_tokens": 773,
            "reasoning_output_tokens": 279
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41563,
            "output_tokens": 124,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42006,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42092,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42285,
            "output_tokens": 303,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42953,
            "output_tokens": 140,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 43286,
            "output_tokens": 605,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 44164,
            "output_tokens": 106,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44590,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44676,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 44868,
            "output_tokens": 628,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45683,
            "output_tokens": 323,
            "reasoning_output_tokens": 231
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 46324,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46406,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46545,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 46593,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 46675,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 47082,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47274,
            "output_tokens": 64,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47400,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47360,
            "input_tokens": 47468,
            "output_tokens": 504,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47872,
            "input_tokens": 47994,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 74,
        "model_tool_calls_by_name": {
          "read": 15,
          "ls": 1,
          "bash": 1,
          "exec": 27,
          "devices": 1,
          "device": 18,
          "write": 3,
          "device_stop": 2,
          "edit": 3,
          "relative_pose": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 397 rosbridge control frames; 25 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/workspace/.kinex/memos/robotwin.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_chunk.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-28-20261113/workspace/.kinex/tools/policy_chunk.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 147,
        "observed_images": 30
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 25,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 160,
        "published_frames": 160,
        "decode_ok": true,
        "duration_s": 4.0,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "7a6f4e389738cd20c90a2f937cc1a462da0d1c67d3357d7da82eda9c4a4a6198",
        "source_sha256": {
          "third_person": "f9eb408c47be53ea538388b66e260b0538eea56a084724f37d68b506f31a3034",
          "left_wrist": "4bbe787ed67137e5e52bf394c7defc9443ed0d6cf3f068d961f02bbe7b6d725c",
          "right_wrist": "262b35604724dd69e08d976e9ebeb9b82142dd3037d5eb9d4da15a22ac2eb8f8"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "a1fc91152795",
          "task": "robotwin/place-a2b-right-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 160,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 160,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15880000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 160,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 160,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15880000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 160,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 160,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15880000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-29-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/29",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "29",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 19.400000000000002,
        "step_limit": 700,
        "steps": 485,
        "success": true,
        "valid": true
      },
      "steps": 485,
      "simulation_time_s": 19.400000000000002,
      "wall_time_s": 2052.059579,
      "timing": {
        "simulation_time_s": 19.400000000000002,
        "wall_time_s": 2052.059579,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move both the soft bread with golden edges and the soft rectangular bread loaf inside the white basket with beige bottom.",
      "native_instruction": "Move both the soft bread with golden edges and the soft rectangular bread loaf inside the white basket with beige bottom.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9708241579726363,
        "cache_read_input_tokens": 1075712,
        "cache_reported_input_tokens": 1108040,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1108040,
        "cached_input_tokens": 1075712,
        "cost_usd": null,
        "fresh_input_tokens": 32328,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1108040,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1075712,
        "known_fresh_input_tokens": 32328,
        "known_input_tokens": 1108040,
        "known_output_tokens": 5579,
        "known_reasoning_output_tokens": 902,
        "output_tokens": 5579,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 902,
        "reasoning_reported_output_tokens": 5579,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 42,
          "cache_write_input_tokens": 42,
          "cache_write_reported_input_tokens": 42,
          "cached_input_tokens": 42,
          "fresh_input_tokens": 42,
          "input_tokens": 42,
          "output_tokens": 42,
          "reasoning_output_tokens": 42,
          "reasoning_reported_output_tokens": 42
        },
        "request_attempts": 42,
        "response_count": 42,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1113619,
        "uncached_input_tokens": 32328,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11826,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11866,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11999,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12049,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 14448,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14601,
            "output_tokens": 177,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14592,
            "input_tokens": 18069,
            "output_tokens": 199,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 21911,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22237,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 23211,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23402,
            "output_tokens": 73,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23844,
            "output_tokens": 551,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24418,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24632,
            "output_tokens": 128,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24789,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25634,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25722,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25961,
            "output_tokens": 290,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26280,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26408,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26500,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 27092,
            "output_tokens": 86,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27525,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27943,
            "output_tokens": 100,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28253,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28711,
            "output_tokens": 1029,
            "reasoning_output_tokens": 300
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29763,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30207,
            "output_tokens": 98,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30674,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31119,
            "output_tokens": 361,
            "reasoning_output_tokens": 100
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31888,
            "output_tokens": 187,
            "reasoning_output_tokens": 122
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32366,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32717,
            "output_tokens": 82,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33095,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33446,
            "output_tokens": 184,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33922,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34274,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34624,
            "output_tokens": 78,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34848,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 34941,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35133,
            "output_tokens": 534,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35692,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 49,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 14,
          "find": 1,
          "devices": 3,
          "exec": 19,
          "device": 2,
          "write": 3,
          "edit": 2,
          "device_stop": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 485 rosbridge control frames; 39 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/bread-basket.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/workspace/.kinex/memos/bread-basket.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/run_vla.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/workspace/.kinex/tools/run_vla.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-29-20261113/workspace/.kinex/tools/tcp_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 94,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 39,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 195,
        "published_frames": 195,
        "decode_ok": true,
        "duration_s": 4.875,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "31f664a3e078feaa7986bcc1740b9c47615636d198d3014efe7876914c2cda81",
        "source_sha256": {
          "third_person": "69aa22dad90d5bf285871a7a403a5666730d1814faa0adef62e0779df716ac26",
          "left_wrist": "7979947712769d34ae4aa9cfcbf0bd43439c034b370f6189f326de50e2e9c906",
          "right_wrist": "b91034467784a40903ff005103d97ef72a48807c16ea783bee47d752ac77abac"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "199de5f70bcc",
          "task": "robotwin/place-bread-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 196,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 195,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19400000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 196,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 195,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19400000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 196,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 195,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19400000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-30-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/30",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "30",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 15.32,
        "step_limit": 500,
        "steps": 383,
        "success": true,
        "valid": true
      },
      "steps": 383,
      "simulation_time_s": 15.32,
      "wall_time_s": 1508.127575,
      "timing": {
        "simulation_time_s": 15.32,
        "wall_time_s": 1508.127575,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move the spongy golden brown bread from the table into the versatile skillet for frying and sauteing.",
      "native_instruction": "Move the spongy golden brown bread from the table into the versatile skillet for frying and sauteing.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9728023243612137,
        "cache_read_input_tokens": 1307136,
        "cache_reported_input_tokens": 1343681,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1343681,
        "cached_input_tokens": 1307136,
        "cost_usd": null,
        "fresh_input_tokens": 36545,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1343681,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1307136,
        "known_fresh_input_tokens": 36545,
        "known_input_tokens": 1343681,
        "known_output_tokens": 6351,
        "known_reasoning_output_tokens": 1050,
        "output_tokens": 6351,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1050,
        "reasoning_reported_output_tokens": 6351,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 43,
          "cache_write_input_tokens": 43,
          "cache_write_reported_input_tokens": 43,
          "cached_input_tokens": 43,
          "fresh_input_tokens": 43,
          "input_tokens": 43,
          "output_tokens": 43,
          "reasoning_output_tokens": 43,
          "reasoning_reported_output_tokens": 43
        },
        "request_attempts": 43,
        "response_count": 43,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1350032,
        "uncached_input_tokens": 36545,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11825,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11865,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11998,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12046,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13959,
            "output_tokens": 122,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13952,
            "input_tokens": 14291,
            "output_tokens": 226,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14336,
            "input_tokens": 19382,
            "output_tokens": 131,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 21582,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 22586,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22912,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23103,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23511,
            "output_tokens": 652,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24184,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 25026,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25116,
            "output_tokens": 52,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 26429,
            "output_tokens": 282,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26738,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26949,
            "output_tokens": 124,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27100,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 28130,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 30236,
            "output_tokens": 80,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 32350,
            "output_tokens": 72,
            "reasoning_output_tokens": 6
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 34147,
            "output_tokens": 90,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 35977,
            "output_tokens": 52,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36091,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36343,
            "output_tokens": 182,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36780,
            "output_tokens": 335,
            "reasoning_output_tokens": 243
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 38758,
            "output_tokens": 868,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39647,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40081,
            "output_tokens": 246,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40803,
            "output_tokens": 178,
            "reasoning_output_tokens": 100
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41345,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41691,
            "output_tokens": 55,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41958,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42236,
            "output_tokens": 73,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42594,
            "output_tokens": 101,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43390,
            "output_tokens": 591,
            "reasoning_output_tokens": 243
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44440,
            "output_tokens": 217,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 44799,
            "output_tokens": 47,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 44908,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45101,
            "output_tokens": 68,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45339,
            "output_tokens": 573,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 45935,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 55,
        "model_tool_calls_by_name": {
          "ls": 5,
          "read": 18,
          "find": 2,
          "exec": 18,
          "devices": 1,
          "device": 3,
          "write": 3,
          "device_stop": 3,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 383 rosbridge control frames; 49 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-control.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/workspace/.kinex/memos/robotwin-control.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-30-20261113/workspace/.kinex/tools/policy_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 101,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 49,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 154,
        "published_frames": 154,
        "decode_ok": true,
        "duration_s": 3.85,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f34fcbfcc89caf8918b7092ecf8d3358bfdc82079311610307f234fea92ee7b6",
        "source_sha256": {
          "third_person": "ceab530ab0c8eed875edb0823aa30f8366753230e28687908a01dcce2ab84cbb",
          "left_wrist": "273abca20df78a9b310c345aa9d54143bf642e0d34c850a1dc5469fecf27d7ff",
          "right_wrist": "a82c37d847fb92ef52c45d4bdd43c09d430b90fc0172c4ac8303cbf3397ff16b"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "b7e56041a5a9",
          "task": "robotwin/place-bread-skillet-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 155,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 154,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15320000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 155,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 154,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15320000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 155,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 154,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15320000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-31-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/31",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "31",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.08,
        "step_limit": 500,
        "steps": 302,
        "success": true,
        "valid": true
      },
      "steps": 302,
      "simulation_time_s": 12.08,
      "wall_time_s": 1643.425105,
      "timing": {
        "simulation_time_s": 12.08,
        "wall_time_s": 1643.425105,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the compact hamburg box and the Crispy potato fries with red box, place them on the green tray.",
      "native_instruction": "Pick the compact hamburg box and the Crispy potato fries with red box, place them on the green tray.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9731641853531454,
        "cache_read_input_tokens": 1148288,
        "cache_reported_input_tokens": 1179953,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1179953,
        "cached_input_tokens": 1148288,
        "cost_usd": null,
        "fresh_input_tokens": 31665,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1179953,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1148288,
        "known_fresh_input_tokens": 31665,
        "known_input_tokens": 1179953,
        "known_output_tokens": 5298,
        "known_reasoning_output_tokens": 501,
        "output_tokens": 5298,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 501,
        "reasoning_reported_output_tokens": 5298,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 39,
          "cache_write_input_tokens": 39,
          "cache_write_reported_input_tokens": 39,
          "cached_input_tokens": 39,
          "fresh_input_tokens": 39,
          "input_tokens": 39,
          "output_tokens": 39,
          "reasoning_output_tokens": 39,
          "reasoning_reported_output_tokens": 39
        },
        "request_attempts": 39,
        "response_count": 39,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1185251,
        "uncached_input_tokens": 31665,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11825,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11958,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12008,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12297,
            "output_tokens": 252,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 17533,
            "output_tokens": 199,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 22559,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22885,
            "output_tokens": 132,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 24462,
            "output_tokens": 659,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25142,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25984,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26073,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26471,
            "output_tokens": 332,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26830,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27041,
            "output_tokens": 41,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27144,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 28911,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29035,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29987,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30933,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31958,
            "output_tokens": 76,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32993,
            "output_tokens": 83,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 34037,
            "output_tokens": 61,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34160,
            "output_tokens": 170,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 35380,
            "output_tokens": 294,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35985,
            "output_tokens": 159,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36209,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36291,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36629,
            "output_tokens": 599,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37249,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37681,
            "output_tokens": 156,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38197,
            "output_tokens": 109,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38672,
            "output_tokens": 95,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 38829,
            "output_tokens": 87,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 38981,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39030,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39124,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39500,
            "output_tokens": 163,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39690,
            "output_tokens": 565,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40280,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 54,
        "model_tool_calls_by_name": {
          "read": 16,
          "ls": 1,
          "bash": 1,
          "exec": 13,
          "devices": 6,
          "device": 4,
          "write": 3,
          "device_stop": 5,
          "edit": 3,
          "relative_pose": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 302 rosbridge control frames; 46 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-food-tray.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/workspace/.kinex/memos/robotwin-food-tray.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_segment.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/workspace/.kinex/tools/cartesian_segment.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_segment.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-31-20261113/workspace/.kinex/tools/policy_segment.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 96,
        "observed_images": 24
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 46,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 122,
        "published_frames": 122,
        "decode_ok": true,
        "duration_s": 3.05,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "833442f720477fb0d0d05c4387fa9bd13fca458f1caac9ccd303629098c4d220",
        "source_sha256": {
          "third_person": "a2410701803817eeb65112e026bdfa8de1908bd7916c5af5a610b136d77de75c",
          "left_wrist": "725e6126326162e94210dc7f64fa6750322f7fc9c41b9c87f7e0501c7f43660e",
          "right_wrist": "1a36c00d72a3cde7004137a3d32321e28561bcfe16fdd5d6bad9b1fc5a50a65b"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "c4174909f0d3",
          "task": "robotwin/place-burger-fries-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 122,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 122,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12080000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 122,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 122,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12080000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 122,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 122,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12080000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-32-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/32",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "32",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 20.84,
        "step_limit": 700,
        "steps": 521,
        "success": true,
        "valid": true
      },
      "steps": 521,
      "simulation_time_s": 20.84,
      "wall_time_s": 2140.804326,
      "timing": {
        "simulation_time_s": 20.84,
        "wall_time_s": 2140.804326,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to put the red can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "native_instruction": "Use the left arm to put the red can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9651381381857376,
        "cache_read_input_tokens": 1826688,
        "cache_reported_input_tokens": 1892670,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1892670,
        "cached_input_tokens": 1826688,
        "cost_usd": null,
        "fresh_input_tokens": 65982,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1892670,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1826688,
        "known_fresh_input_tokens": 65982,
        "known_input_tokens": 1892670,
        "known_output_tokens": 9487,
        "known_reasoning_output_tokens": 1612,
        "output_tokens": 9487,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1612,
        "reasoning_reported_output_tokens": 9487,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 54,
          "cache_write_input_tokens": 54,
          "cache_write_reported_input_tokens": 54,
          "cached_input_tokens": 54,
          "fresh_input_tokens": 54,
          "input_tokens": 54,
          "output_tokens": 54,
          "reasoning_output_tokens": 54,
          "reasoning_reported_output_tokens": 54
        },
        "request_attempts": 55,
        "response_count": 54,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1902157,
        "uncached_input_tokens": 65982,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11830,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11870,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12003,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12343,
            "output_tokens": 252,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 17579,
            "output_tokens": 168,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 19861,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 22696,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 23860,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 24186,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24378,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24786,
            "output_tokens": 854,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25661,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25880,
            "output_tokens": 125,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26032,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 26921,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27010,
            "output_tokens": 73,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27622,
            "output_tokens": 741,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28390,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 29545,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30709,
            "output_tokens": 155,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 32231,
            "output_tokens": 312,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33341,
            "output_tokens": 290,
            "reasoning_output_tokens": 171
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33776,
            "output_tokens": 1102,
            "reasoning_output_tokens": 138
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34899,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35627,
            "output_tokens": 75,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35729,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36352,
            "output_tokens": 270,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36649,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37248,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37803,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38579,
            "output_tokens": 56,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 40595,
            "output_tokens": 117,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40739,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 41560,
            "output_tokens": 143,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 42202,
            "output_tokens": 221,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42483,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42572,
            "output_tokens": 109,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 43052,
            "output_tokens": 163,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43242,
            "output_tokens": 103,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43844,
            "output_tokens": 125,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44031,
            "output_tokens": 83,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44437,
            "output_tokens": 71,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 45003,
            "output_tokens": 432,
            "reasoning_output_tokens": 188
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45456,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 46051,
            "output_tokens": 131,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46610,
            "output_tokens": 340,
            "reasoning_output_tokens": 168
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47697,
            "output_tokens": 172,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 48375,
            "output_tokens": 112,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 48985,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49200,
            "output_tokens": 80,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49152,
            "input_tokens": 49707,
            "output_tokens": 168,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49792,
            "input_tokens": 50010,
            "output_tokens": 336,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50304,
            "input_tokens": 50373,
            "output_tokens": 651,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50944,
            "input_tokens": 51050,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 65,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 18,
          "bash": 1,
          "exec": 25,
          "devices": 5,
          "device": 1,
          "write": 4,
          "edit": 7,
          "device_stop": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 521 rosbridge control frames; 42 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-place-can-basket.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/workspace/.kinex/memos/robotwin-place-can-basket.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/probe_pixels.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/workspace/.kinex/tools/probe_pixels.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/run_visual_policy.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-32-20261113/workspace/.kinex/tools/run_visual_policy.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 123,
        "observed_images": 20
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 42,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 210,
        "published_frames": 210,
        "decode_ok": true,
        "duration_s": 5.25,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "3edcde6e090708746cc561cb16fd3d208297ccf7b319162c7b327e86ae9fbeab",
        "source_sha256": {
          "third_person": "d761838cad9631a2bd7cb60cddfde450f833c2c1b56e01987c8c461683cf7fe2",
          "left_wrist": "607167b83cab8476721c5ee1bf8a7e4def44ab50bb2dcf6f00a5e1d41806d2ed",
          "right_wrist": "5c467e099cb2a20de8c6bb616f57981d0a4f381cb906541b8c5455359ccdcfdb"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "264d1f662a56",
          "task": "robotwin/place-can-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 210,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 210,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 20840000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 210,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 210,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 20840000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 210,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 210,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 20840000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-33-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/33",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "33",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 28.92,
        "step_limit": 800,
        "steps": 723,
        "success": true,
        "valid": true
      },
      "steps": 723,
      "simulation_time_s": 28.92,
      "wall_time_s": 2420.881405,
      "timing": {
        "simulation_time_s": 28.92,
        "wall_time_s": 2420.881405,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the small red soda can and the lightweight cylindrical metal can in parallel, then put both into the rectangular plasticbox.",
      "native_instruction": "Grab the small red soda can and the lightweight cylindrical metal can in parallel, then put both into the rectangular plasticbox.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9814917833140383,
        "cache_read_input_tokens": 2284800,
        "cache_reported_input_tokens": 2327885,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2327885,
        "cached_input_tokens": 2284800,
        "cost_usd": null,
        "fresh_input_tokens": 43085,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2327885,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2284800,
        "known_fresh_input_tokens": 43085,
        "known_input_tokens": 2327885,
        "known_output_tokens": 9208,
        "known_reasoning_output_tokens": 2372,
        "output_tokens": 9208,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2372,
        "reasoning_reported_output_tokens": 9208,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 64,
          "cache_write_input_tokens": 64,
          "cache_write_reported_input_tokens": 64,
          "cached_input_tokens": 64,
          "fresh_input_tokens": 64,
          "input_tokens": 64,
          "output_tokens": 64,
          "reasoning_output_tokens": 64,
          "reasoning_reported_output_tokens": 64
        },
        "request_attempts": 64,
        "response_count": 64,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2337093,
        "uncached_input_tokens": 43085,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11827,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11867,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12000,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12048,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14048,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13952,
            "input_tokens": 14330,
            "output_tokens": 252,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 19566,
            "output_tokens": 138,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 21773,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 24608,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25582,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25908,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26099,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26507,
            "output_tokens": 731,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27259,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 28104,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28193,
            "output_tokens": 66,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28610,
            "output_tokens": 321,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 28958,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29170,
            "output_tokens": 108,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29305,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 30112,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30933,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31752,
            "output_tokens": 42,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31856,
            "output_tokens": 146,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32844,
            "output_tokens": 338,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33685,
            "output_tokens": 265,
            "reasoning_output_tokens": 223
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 35129,
            "output_tokens": 763,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 35913,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36377,
            "output_tokens": 199,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36757,
            "output_tokens": 321,
            "reasoning_output_tokens": 172
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37464,
            "output_tokens": 170,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38019,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38447,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38884,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39339,
            "output_tokens": 371,
            "reasoning_output_tokens": 297
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40406,
            "output_tokens": 233,
            "reasoning_output_tokens": 159
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40699,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40790,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41387,
            "output_tokens": 165,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41763,
            "output_tokens": 113,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42255,
            "output_tokens": 147,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42788,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43223,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43662,
            "output_tokens": 116,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44201,
            "output_tokens": 133,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44715,
            "output_tokens": 145,
            "reasoning_output_tokens": 120
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45028,
            "output_tokens": 254,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45728,
            "output_tokens": 430,
            "reasoning_output_tokens": 349
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45952,
            "input_tokens": 46854,
            "output_tokens": 224,
            "reasoning_output_tokens": 116
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47144,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 47618,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 48096,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48000,
            "input_tokens": 48531,
            "output_tokens": 78,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48512,
            "input_tokens": 48995,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49440,
            "output_tokens": 116,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49408,
            "input_tokens": 49933,
            "output_tokens": 176,
            "reasoning_output_tokens": 112
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49920,
            "input_tokens": 50487,
            "output_tokens": 80,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50432,
            "input_tokens": 50956,
            "output_tokens": 79,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50816,
            "input_tokens": 51177,
            "output_tokens": 48,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51072,
            "input_tokens": 51287,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51072,
            "input_tokens": 51479,
            "output_tokens": 253,
            "reasoning_output_tokens": 136
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51712,
            "input_tokens": 51759,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51712,
            "input_tokens": 51827,
            "output_tokens": 535,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52224,
            "input_tokens": 52384,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 74,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "find": 1,
          "bash": 1,
          "exec": 36,
          "devices": 3,
          "device": 4,
          "write": 3,
          "device_stop": 4,
          "edit": 4
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 723 rosbridge control frames; 55 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/place-cans.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/workspace/.kinex/memos/place-cans.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_pair.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/workspace/.kinex/tools/cartesian_pair.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_batch.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-33-20261113/workspace/.kinex/tools/policy_batch.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 141,
        "observed_images": 36
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 55,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 290,
        "published_frames": 290,
        "decode_ok": true,
        "duration_s": 7.25,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "05a10b962177b098d4282dbc4d7958b7eadd7b56d8d6e8ca3cdfb374af4fbdb4",
        "source_sha256": {
          "third_person": "36873ce27985cca87ae925cfec8f4d8f26b77b55afcdb705597e3586eed07588",
          "left_wrist": "0f4cb79bba1db95d8583af447af8a7e9a9e0d5300022b6bba2953cc7b8a5aaf8",
          "right_wrist": "ecb7301f315ebd6d5d1e0e43f394bb9798772af449c6da81dedc04dee293b11a"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "2de8e9920a09",
          "task": "robotwin/place-cans-plasticbox-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 291,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 290,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28920000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 291,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 290,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28920000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 291,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 290,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28920000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-34-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/34",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "34",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.96,
        "step_limit": 400,
        "steps": 324,
        "success": true,
        "valid": true
      },
      "steps": 324,
      "simulation_time_s": 12.96,
      "wall_time_s": 1881.749081,
      "timing": {
        "simulation_time_s": 12.96,
        "wall_time_s": 1881.749081,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the ceramic drinking cup with floral design on the gray plate for dining.",
      "native_instruction": "Place the ceramic drinking cup with floral design on the gray plate for dining.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9773387237885401,
        "cache_read_input_tokens": 1491328,
        "cache_reported_input_tokens": 1525907,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1525907,
        "cached_input_tokens": 1491328,
        "cost_usd": null,
        "fresh_input_tokens": 34579,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1525907,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1491328,
        "known_fresh_input_tokens": 34579,
        "known_input_tokens": 1525907,
        "known_output_tokens": 6562,
        "known_reasoning_output_tokens": 1042,
        "output_tokens": 6562,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1042,
        "reasoning_reported_output_tokens": 6562,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 48,
          "cache_write_input_tokens": 48,
          "cache_write_reported_input_tokens": 48,
          "cached_input_tokens": 48,
          "fresh_input_tokens": 48,
          "input_tokens": 48,
          "output_tokens": 48,
          "reasoning_output_tokens": 48,
          "reasoning_reported_output_tokens": 48
        },
        "request_attempts": 48,
        "response_count": 48,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1532469,
        "uncached_input_tokens": 34579,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11859,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11992,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14371,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 17974,
            "output_tokens": 208,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 21849,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 23392,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23718,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24692,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24883,
            "output_tokens": 77,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25329,
            "output_tokens": 556,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25906,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26111,
            "output_tokens": 124,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26262,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 27098,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27184,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27602,
            "output_tokens": 57,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 29612,
            "output_tokens": 636,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30269,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30893,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31514,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31632,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31721,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32028,
            "output_tokens": 73,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32662,
            "output_tokens": 91,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32963,
            "output_tokens": 288,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33804,
            "output_tokens": 114,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34480,
            "output_tokens": 358,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35666,
            "output_tokens": 112,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36244,
            "output_tokens": 815,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37080,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37765,
            "output_tokens": 78,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 37870,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38398,
            "output_tokens": 92,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38959,
            "output_tokens": 63,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39044,
            "output_tokens": 297,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39602,
            "output_tokens": 102,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40172,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40701,
            "output_tokens": 199,
            "reasoning_output_tokens": 136
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 40960,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41047,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41534,
            "output_tokens": 61,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41701,
            "output_tokens": 154,
            "reasoning_output_tokens": 92
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42327,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42845,
            "output_tokens": 83,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43070,
            "output_tokens": 52,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 43354,
            "output_tokens": 572,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 43949,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 58,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 16,
          "find": 1,
          "exec": 21,
          "devices": 5,
          "device": 1,
          "write": 4,
          "edit": 2,
          "device_stop": 4,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 324 rosbridge control frames; 41 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-control.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/workspace/.kinex/memos/robotwin-control.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/workspace/.kinex/tools/cartesian_step.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vla_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-34-20261113/workspace/.kinex/tools/vla_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 109,
        "observed_images": 29
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 41,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 131,
        "published_frames": 131,
        "decode_ok": true,
        "duration_s": 3.275,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "6ba502112ba8e6b28f9fad7a975a5c7e5e7a22d2385e4bc2620eef9d45f71886",
        "source_sha256": {
          "third_person": "3ded6383cb8affa0a75711763e8c54e0fded1c8ef41ef299dc54dce387768fcd",
          "left_wrist": "17e03ac88e4e542e54b6e33d7ab753872a2b1d5e5be16a8110f7036c82171a61",
          "right_wrist": "3fa7080d0e390eeea6c184009e450c8556b13e58d7640dfb1c5307056e894721"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "60e55dc13e9b",
          "task": "robotwin/place-container-plate-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 131,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 131,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 131,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 131,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 131,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 131,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-35-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/35",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "35",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 17.2,
        "step_limit": 500,
        "steps": 430,
        "success": true,
        "valid": true
      },
      "steps": 430,
      "simulation_time_s": 17.2,
      "wall_time_s": 1910.948552,
      "timing": {
        "simulation_time_s": 17.2,
        "wall_time_s": 1910.948552,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the small smooth cup onto the light gray coaster gently.",
      "native_instruction": "Place the small smooth cup onto the light gray coaster gently.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9759572427137402,
        "cache_read_input_tokens": 1425408,
        "cache_reported_input_tokens": 1460523,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1460523,
        "cached_input_tokens": 1425408,
        "cost_usd": null,
        "fresh_input_tokens": 35115,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1460523,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1425408,
        "known_fresh_input_tokens": 35115,
        "known_input_tokens": 1460523,
        "known_output_tokens": 7953,
        "known_reasoning_output_tokens": 1368,
        "output_tokens": 7953,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1368,
        "reasoning_reported_output_tokens": 7953,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 45,
          "cache_write_input_tokens": 45,
          "cache_write_reported_input_tokens": 45,
          "cached_input_tokens": 45,
          "fresh_input_tokens": 45,
          "input_tokens": 45,
          "output_tokens": 45,
          "reasoning_output_tokens": 45,
          "reasoning_reported_output_tokens": 45
        },
        "request_attempts": 45,
        "response_count": 45,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1468476,
        "uncached_input_tokens": 35115,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11814,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11854,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11987,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14994,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 15163,
            "output_tokens": 222,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15232,
            "input_tokens": 19577,
            "output_tokens": 220,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 24142,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24468,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24660,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 26077,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26485,
            "output_tokens": 702,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27207,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27410,
            "output_tokens": 109,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27545,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28413,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28503,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28923,
            "output_tokens": 69,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 30220,
            "output_tokens": 737,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30977,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31797,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31912,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 31999,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32402,
            "output_tokens": 84,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33246,
            "output_tokens": 89,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 34091,
            "output_tokens": 166,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 35035,
            "output_tokens": 270,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35888,
            "output_tokens": 412,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36556,
            "output_tokens": 269,
            "reasoning_output_tokens": 227
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 38004,
            "output_tokens": 879,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 38904,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39472,
            "output_tokens": 127,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39626,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40370,
            "output_tokens": 297,
            "reasoning_output_tokens": 206
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 41013,
            "output_tokens": 107,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41471,
            "output_tokens": 138,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41691,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42127,
            "output_tokens": 201,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42675,
            "output_tokens": 413,
            "reasoning_output_tokens": 120
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43289,
            "output_tokens": 192,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43828,
            "output_tokens": 129,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 44303,
            "output_tokens": 237,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44682,
            "output_tokens": 59,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 44803,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45070,
            "output_tokens": 756,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 45850,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 56,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 17,
          "find": 1,
          "bash": 1,
          "exec": 21,
          "devices": 4,
          "device": 1,
          "write": 4,
          "edit": 2,
          "device_stop": 3,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 430 rosbridge control frames; 53 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-cup.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/workspace/.kinex/memos/robotwin-cup.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/run_policy.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-35-20261113/workspace/.kinex/tools/run_policy.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 104,
        "observed_images": 20
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 53,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 173,
        "published_frames": 173,
        "decode_ok": true,
        "duration_s": 4.325,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "fccc4efdfd67e37bed75f904c7f0ab69d5afe75f0f8b12ed4f345702fca8018b",
        "source_sha256": {
          "third_person": "fedf0236875a0723029dc1e54cfc2f44edafb1695f398ae0f3aeebd54091dccf",
          "left_wrist": "e582344dc8a4f3ea006590ca5b452c8bd760a4b897d5ff25f800f0471f57eb75",
          "right_wrist": "8a29b9e9d431529510f5663639966776159fd8b9dc035d031eed2f5df2ec21fe"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "9d026d77a5bc",
          "task": "robotwin/place-empty-cup-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 174,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 173,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 174,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 173,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 174,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 173,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-36-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/36",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "36",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 13.16,
        "step_limit": 400,
        "steps": 329,
        "success": true,
        "valid": true
      },
      "steps": 329,
      "simulation_time_s": 13.16,
      "wall_time_s": 1594.675523,
      "timing": {
        "simulation_time_s": 13.16,
        "wall_time_s": 1594.675523,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the beige circular blade fan, place it on the Pink mat, face it toward the robot",
      "native_instruction": "Pick the beige circular blade fan, place it on the Pink mat, face it toward the robot",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9713561290436972,
        "cache_read_input_tokens": 1095680,
        "cache_reported_input_tokens": 1127990,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1127990,
        "cached_input_tokens": 1095680,
        "cost_usd": null,
        "fresh_input_tokens": 32310,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1127990,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1095680,
        "known_fresh_input_tokens": 32310,
        "known_input_tokens": 1127990,
        "known_output_tokens": 5974,
        "known_reasoning_output_tokens": 1091,
        "output_tokens": 5974,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1091,
        "reasoning_reported_output_tokens": 5974,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 41,
          "cache_write_input_tokens": 41,
          "cache_write_reported_input_tokens": 41,
          "cached_input_tokens": 41,
          "fresh_input_tokens": 41,
          "input_tokens": 41,
          "output_tokens": 41,
          "reasoning_output_tokens": 41,
          "reasoning_reported_output_tokens": 41
        },
        "request_attempts": 41,
        "response_count": 41,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1133964,
        "uncached_input_tokens": 32310,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11859,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11992,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 12042,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13821,
            "output_tokens": 132,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13824,
            "input_tokens": 14163,
            "output_tokens": 208,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14208,
            "input_tokens": 18444,
            "output_tokens": 170,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 21475,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 21801,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22775,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 22967,
            "output_tokens": 62,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23398,
            "output_tokens": 588,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24010,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24857,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 24944,
            "output_tokens": 71,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25516,
            "output_tokens": 336,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25882,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26096,
            "output_tokens": 128,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26254,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26894,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27529,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28167,
            "output_tokens": 91,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28827,
            "output_tokens": 181,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 30078,
            "output_tokens": 355,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30971,
            "output_tokens": 105,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31542,
            "output_tokens": 201,
            "reasoning_output_tokens": 159
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32922,
            "output_tokens": 44,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33028,
            "output_tokens": 615,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33666,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34004,
            "output_tokens": 121,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34479,
            "output_tokens": 111,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34944,
            "output_tokens": 66,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35371,
            "output_tokens": 109,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35838,
            "output_tokens": 296,
            "reasoning_output_tokens": 157
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36491,
            "output_tokens": 216,
            "reasoning_output_tokens": 127
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37065,
            "output_tokens": 212,
            "reasoning_output_tokens": 173
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37602,
            "output_tokens": 136,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38098,
            "output_tokens": 75,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38315,
            "output_tokens": 69,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38780,
            "output_tokens": 460,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39264,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 54,
        "model_tool_calls_by_name": {
          "ls": 5,
          "read": 17,
          "find": 2,
          "devices": 4,
          "exec": 17,
          "device": 2,
          "write": 3,
          "device_stop": 2,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 329 rosbridge control frames; 40 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/workspace/.kinex/memos/robotwin.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_rollout.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-36-20261113/workspace/.kinex/tools/policy_rollout.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 98,
        "observed_images": 27
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 40,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 133,
        "published_frames": 133,
        "decode_ok": true,
        "duration_s": 3.325,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "40a9f9d6c0f3850806fc204933608918af3495be3028acb52d5ce15a510f349a",
        "source_sha256": {
          "third_person": "5dac8efce206fb11e9c8495797e64dbf3052d7a832624aff94493f9842b80986",
          "left_wrist": "229f3fbd149ce0a335664c2644ceb284e5ad21792818f0e682c59b033ae0fb5e",
          "right_wrist": "137c885393f2f4737eb7b055a36bd4a589f2179e3eb6f0e5147fcd0f66db9ccb"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "223a5a4f4065",
          "task": "robotwin/place-fan-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 133,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 133,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13160000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 133,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 133,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13160000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 133,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 133,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13160000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-37-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/37",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "37",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 2
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.64,
        "step_limit": 400,
        "steps": 316,
        "success": true,
        "valid": true
      },
      "steps": 316,
      "simulation_time_s": 12.64,
      "wall_time_s": 1653.160114,
      "timing": {
        "simulation_time_s": 12.64,
        "wall_time_s": 1653.160114,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the dark gray computer mouse with logo with the left arm and drop it onto the Cyan mat.",
      "native_instruction": "Pick up the dark gray computer mouse with logo with the left arm and drop it onto the Cyan mat.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9736756963440101,
        "cache_read_input_tokens": 1324160,
        "cache_reported_input_tokens": 1359960,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1359960,
        "cached_input_tokens": 1324160,
        "cost_usd": null,
        "fresh_input_tokens": 35800,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1359960,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1324160,
        "known_fresh_input_tokens": 35800,
        "known_input_tokens": 1359960,
        "known_output_tokens": 5793,
        "known_reasoning_output_tokens": 487,
        "output_tokens": 5793,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 487,
        "reasoning_reported_output_tokens": 5793,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 46,
          "cache_write_input_tokens": 46,
          "cache_write_reported_input_tokens": 46,
          "cached_input_tokens": 46,
          "fresh_input_tokens": 46,
          "input_tokens": 46,
          "output_tokens": 46,
          "reasoning_output_tokens": 46,
          "reasoning_reported_output_tokens": 46
        },
        "request_attempts": 46,
        "response_count": 46,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1365753,
        "uncached_input_tokens": 35800,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11822,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11862,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11995,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12043,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12332,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12483,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 14195,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14080,
            "input_tokens": 14341,
            "output_tokens": 285,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 21029,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21599,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 21925,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 23342,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23532,
            "output_tokens": 59,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23960,
            "output_tokens": 557,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24539,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24752,
            "output_tokens": 108,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 24888,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25748,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25837,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26256,
            "output_tokens": 60,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 28355,
            "output_tokens": 645,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29028,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29618,
            "output_tokens": 44,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29724,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29810,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30002,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 32075,
            "output_tokens": 76,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 34181,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 36293,
            "output_tokens": 78,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36581,
            "output_tokens": 304,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37343,
            "output_tokens": 73,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37882,
            "output_tokens": 753,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38656,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39121,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39593,
            "output_tokens": 180,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39974,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40447,
            "output_tokens": 238,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 41085,
            "output_tokens": 107,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41972,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42063,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42371,
            "output_tokens": 147,
            "reasoning_output_tokens": 6
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42660,
            "output_tokens": 45,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 42767,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 42960,
            "output_tokens": 203,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 43190,
            "output_tokens": 517,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 43729,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 53,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "bash": 2,
          "exec": 16,
          "devices": 3,
          "device": 3,
          "write": 3,
          "edit": 4,
          "device_stop": 4
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 316 rosbridge control frames; 41 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/workspace/.kinex/memos/robotwin.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vla_rollout.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-37-20261113/workspace/.kinex/tools/vla_rollout.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 102,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 41,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 128,
        "published_frames": 128,
        "decode_ok": true,
        "duration_s": 3.2,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "89de8b8e168c275387b433023760290ae7e631a48c5443cbe4c308feeacadd2e",
        "source_sha256": {
          "third_person": "8745a2f97acc5cca65d8b17146a9f7e62f601b3f9e04555462c17373579ad272",
          "left_wrist": "6cdced4aab70ebfe5e7ae55fda3eec23cd80f10a29cd74c064c45ed5181d3849",
          "right_wrist": "1284e60d6c9f94348eb469fbc54705a2a1a628084f8d3e6537418b851e239eba"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "cd41387a4ca7",
          "task": "robotwin/place-mouse-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 128,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 128,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 128,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 128,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 128,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 128,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-38-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/38",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "38",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 21.04,
        "step_limit": 700,
        "steps": 526,
        "success": false,
        "valid": true
      },
      "steps": 526,
      "simulation_time_s": 21.04,
      "wall_time_s": 1945.1139,
      "timing": {
        "simulation_time_s": 21.04,
        "wall_time_s": 1945.1139,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the toycar with rounded edges and place into the basket with rounded edges.",
      "native_instruction": "Grab the toycar with rounded edges and place into the basket with rounded edges.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9778797343638791,
        "cache_read_input_tokens": 1662464,
        "cache_reported_input_tokens": 1700070,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1700070,
        "cached_input_tokens": 1662464,
        "cost_usd": null,
        "fresh_input_tokens": 37606,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1700070,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1662464,
        "known_fresh_input_tokens": 37606,
        "known_input_tokens": 1700070,
        "known_output_tokens": 7992,
        "known_reasoning_output_tokens": 1362,
        "output_tokens": 7992,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1362,
        "reasoning_reported_output_tokens": 7992,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 54,
          "cache_write_input_tokens": 54,
          "cache_write_reported_input_tokens": 54,
          "cached_input_tokens": 54,
          "fresh_input_tokens": 54,
          "input_tokens": 54,
          "output_tokens": 54,
          "reasoning_output_tokens": 54,
          "reasoning_reported_output_tokens": 54
        },
        "request_attempts": 54,
        "response_count": 54,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1708062,
        "uncached_input_tokens": 37606,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11991,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12041,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13659,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13568,
            "input_tokens": 13745,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13696,
            "input_tokens": 13896,
            "output_tokens": 196,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13952,
            "input_tokens": 18165,
            "output_tokens": 215,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 22399,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22725,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 22916,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23890,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24298,
            "output_tokens": 614,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 24933,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25138,
            "output_tokens": 127,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25292,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 26165,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26254,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26818,
            "output_tokens": 59,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 28830,
            "output_tokens": 675,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29526,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29646,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29736,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29974,
            "output_tokens": 120,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30662,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31303,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31932,
            "output_tokens": 69,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32211,
            "output_tokens": 297,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33099,
            "output_tokens": 429,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33871,
            "output_tokens": 867,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34759,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35449,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35536,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35921,
            "output_tokens": 106,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36342,
            "output_tokens": 58,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36422,
            "output_tokens": 186,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36828,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37580,
            "output_tokens": 89,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38360,
            "output_tokens": 75,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 39609,
            "output_tokens": 71,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40428,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40581,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 41519,
            "output_tokens": 107,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41941,
            "output_tokens": 264,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42598,
            "output_tokens": 161,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43076,
            "output_tokens": 108,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 44065,
            "output_tokens": 92,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44546,
            "output_tokens": 181,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45044,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45106,
            "output_tokens": 220,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45549,
            "output_tokens": 400,
            "reasoning_output_tokens": 295
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46262,
            "output_tokens": 318,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 46607,
            "output_tokens": 521,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47151,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 62,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 17,
          "find": 1,
          "bash": 1,
          "exec": 25,
          "devices": 3,
          "device": 1,
          "write": 4,
          "edit": 4,
          "device_stop": 4
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 526 rosbridge control frames; 50 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-control.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/workspace/.kinex/memos/robotwin-control.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_batch.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-38-20261113/workspace/.kinex/tools/policy_batch.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 119,
        "observed_images": 23
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 50,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 212,
        "published_frames": 212,
        "decode_ok": true,
        "duration_s": 5.3,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "1b31aa8b2288051ae278c02dc300c79fdee2d837df14c7067462784d38cdab00",
        "source_sha256": {
          "third_person": "fbb76e442a4107f92eff7b5ce685242f2ea9a59528c379bc02513d97ab2e72b6",
          "left_wrist": "3caa70a86e73762dfb8e5b0c2fe69bf9688b30cf1053bb7990626aab21919210",
          "right_wrist": "5eba8e87c7c66662c4cb607481835268b7c01938d9a7b307d28b7de41eb82f89"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "f0f36a5558b0",
          "task": "robotwin/place-object-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 212,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 212,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 21040000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 212,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 212,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 21040000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 212,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 212,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 21040000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-39-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/39",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "39",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 13.52,
        "step_limit": 400,
        "steps": 338,
        "success": true,
        "valid": true
      },
      "steps": 338,
      "simulation_time_s": 13.52,
      "wall_time_s": 1295.333541,
      "timing": {
        "simulation_time_s": 13.52,
        "wall_time_s": 1295.333541,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use one arm to pick the palm-sized black stapler for the black screen electronicscale",
      "native_instruction": "Use one arm to pick the palm-sized black stapler for the black screen electronicscale",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9557650556819683,
        "cache_read_input_tokens": 1135360,
        "cache_reported_input_tokens": 1187907,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1187907,
        "cached_input_tokens": 1135360,
        "cost_usd": null,
        "fresh_input_tokens": 52547,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1187907,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1135360,
        "known_fresh_input_tokens": 52547,
        "known_input_tokens": 1187907,
        "known_output_tokens": 5476,
        "known_reasoning_output_tokens": 564,
        "output_tokens": 5476,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 564,
        "reasoning_reported_output_tokens": 5476,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 38,
          "cache_write_input_tokens": 38,
          "cache_write_reported_input_tokens": 38,
          "cached_input_tokens": 38,
          "fresh_input_tokens": 38,
          "input_tokens": 38,
          "output_tokens": 38,
          "reasoning_output_tokens": 38,
          "reasoning_reported_output_tokens": 38
        },
        "request_attempts": 38,
        "response_count": 38,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1193383,
        "uncached_input_tokens": 52547,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11859,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11992,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12332,
            "output_tokens": 252,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 17568,
            "output_tokens": 196,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 21317,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21643,
            "output_tokens": 132,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 23220,
            "output_tokens": 698,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 23941,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24818,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 24906,
            "output_tokens": 227,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25160,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25368,
            "output_tokens": 56,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 27412,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27547,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 31630,
            "output_tokens": 86,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31743,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32578,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33416,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 34254,
            "output_tokens": 84,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 35094,
            "output_tokens": 317,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35955,
            "output_tokens": 113,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36181,
            "output_tokens": 54,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36701,
            "output_tokens": 932,
            "reasoning_output_tokens": 160
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37654,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37813,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38532,
            "output_tokens": 65,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38624,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39239,
            "output_tokens": 140,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39891,
            "output_tokens": 73,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40478,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 41087,
            "output_tokens": 213,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41815,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42422,
            "output_tokens": 74,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42638,
            "output_tokens": 47,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 42747,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43015,
            "output_tokens": 461,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43498,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 50,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 16,
          "bash": 1,
          "exec": 20,
          "devices": 2,
          "device": 1,
          "write": 3,
          "device_stop": 2,
          "edit": 4
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 338 rosbridge control frames; 50 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/workspace/.kinex/memos/robotwin.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_batch.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-39-20261113/workspace/.kinex/tools/policy_batch.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 91,
        "observed_images": 29
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 50,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 136,
        "published_frames": 136,
        "decode_ok": true,
        "duration_s": 3.4,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "b87b46cbc83011a285c842f9a9706772bffad5b02209faac9aeceed7b136474c",
        "source_sha256": {
          "third_person": "029207d2c5f5f6852e6d43e4c49aa1dca6d1c6590ca2cbdb406e70d91ece56b8",
          "left_wrist": "5cac0a12ca68f7637ad07d7fe9b489653d95c8c4d882bd5d0e41f158f744f086",
          "right_wrist": "ce162974994e4cbef49b226febb7b548c2d277b4ccdbdc314a38e0e45ccaf79c"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "59d4e67795c8",
          "task": "robotwin/place-object-scale-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 137,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 136,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13520000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 137,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 136,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13520000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 137,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 136,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13520000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-40-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/40",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "40",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 10.76,
        "step_limit": 400,
        "steps": 269,
        "success": true,
        "valid": true
      },
      "steps": 269,
      "simulation_time_s": 10.76,
      "wall_time_s": 1065.841922,
      "timing": {
        "simulation_time_s": 10.76,
        "wall_time_s": 1065.841922,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set the rectangular remotecontrol on the displaystand with hollow center.",
      "native_instruction": "Set the rectangular remotecontrol on the displaystand with hollow center.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9548734539608733,
        "cache_read_input_tokens": 1048960,
        "cache_reported_input_tokens": 1098533,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1098533,
        "cached_input_tokens": 1048960,
        "cost_usd": null,
        "fresh_input_tokens": 49573,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1098533,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1048960,
        "known_fresh_input_tokens": 49573,
        "known_input_tokens": 1098533,
        "known_output_tokens": 4976,
        "known_reasoning_output_tokens": 647,
        "output_tokens": 4976,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 647,
        "reasoning_reported_output_tokens": 4976,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 39,
          "cache_write_input_tokens": 39,
          "cache_write_reported_input_tokens": 39,
          "cached_input_tokens": 39,
          "fresh_input_tokens": 39,
          "input_tokens": 39,
          "output_tokens": 39,
          "reasoning_output_tokens": 39,
          "reasoning_reported_output_tokens": 39
        },
        "request_attempts": 39,
        "response_count": 39,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1103509,
        "uncached_input_tokens": 49573,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11991,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12331,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 15139,
            "output_tokens": 191,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15232,
            "input_tokens": 18733,
            "output_tokens": 175,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 22558,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22884,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23858,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24050,
            "output_tokens": 67,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24486,
            "output_tokens": 556,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25063,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25898,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25988,
            "output_tokens": 66,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26402,
            "output_tokens": 276,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26705,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26909,
            "output_tokens": 110,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27046,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 28141,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 29241,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 30343,
            "output_tokens": 114,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31339,
            "output_tokens": 123,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31944,
            "output_tokens": 268,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32669,
            "output_tokens": 653,
            "reasoning_output_tokens": 86
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33343,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33686,
            "output_tokens": 47,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33784,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33923,
            "output_tokens": 160,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34350,
            "output_tokens": 91,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34727,
            "output_tokens": 299,
            "reasoning_output_tokens": 237
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35310,
            "output_tokens": 101,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35694,
            "output_tokens": 73,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35818,
            "output_tokens": 117,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36184,
            "output_tokens": 176,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36502,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36595,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36862,
            "output_tokens": 82,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 36971,
            "output_tokens": 397,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37390,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 49,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 14,
          "bash": 1,
          "exec": 16,
          "devices": 3,
          "device": 4,
          "write": 3,
          "device_stop": 2,
          "edit": 3,
          "relative_pose": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 269 rosbridge control frames; 35 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/workspace/.kinex/memos/robotwin.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vla_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-40-20261113/workspace/.kinex/tools/vla_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 91,
        "observed_images": 23
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 35,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 109,
        "published_frames": 109,
        "decode_ok": true,
        "duration_s": 2.725,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "b0ede2d394beca1fdcd0a60a2d5fc5f3977661e56cabeb82afc711687eaea136",
        "source_sha256": {
          "third_person": "d8078379b19c611adcd9e67602a5546e12f5e1c6ebc6abf65da06d9d05c91b17",
          "left_wrist": "9ecfc28792ac0658634ecd70a0ab7cf11f21d043d5378590df00a45c5414f6ca",
          "right_wrist": "cacbc4354e36727fa3868d08b88f8a82481f7503daf730dac9b52f3868f6a2f3"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "0317b88e0afb",
          "task": "robotwin/place-object-stand-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 109,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 109,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10760000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 109,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 109,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10760000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 109,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 109,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10760000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-41-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/41",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "41",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "step_limit",
        "simulation_time_s": 16.0,
        "step_limit": 400,
        "steps": 400,
        "success": false,
        "valid": true
      },
      "steps": 400,
      "simulation_time_s": 16.0,
      "wall_time_s": 2022.783411,
      "timing": {
        "simulation_time_s": 16.0,
        "wall_time_s": 2022.783411,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set the phone with shiny black back on the L-shaped smooth brown phonestand.",
      "native_instruction": "Set the phone with shiny black back on the L-shaped smooth brown phonestand.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9585479513128096,
        "cache_read_input_tokens": 1535104,
        "cache_reported_input_tokens": 1601489,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1601489,
        "cached_input_tokens": 1535104,
        "cost_usd": null,
        "fresh_input_tokens": 66385,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1601489,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1535104,
        "known_fresh_input_tokens": 66385,
        "known_input_tokens": 1601489,
        "known_output_tokens": 8290,
        "known_reasoning_output_tokens": 2026,
        "output_tokens": 8290,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2026,
        "reasoning_reported_output_tokens": 8290,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 51,
          "cache_write_input_tokens": 51,
          "cache_write_reported_input_tokens": 51,
          "cached_input_tokens": 51,
          "fresh_input_tokens": 51,
          "input_tokens": 51,
          "output_tokens": 51,
          "reasoning_output_tokens": 51,
          "reasoning_reported_output_tokens": 51
        },
        "request_attempts": 51,
        "response_count": 51,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1609779,
        "uncached_input_tokens": 66385,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11820,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11860,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11993,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12041,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14120,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14080,
            "input_tokens": 14409,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 18012,
            "output_tokens": 191,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 21846,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22172,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 23589,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23779,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24187,
            "output_tokens": 661,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 24869,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25076,
            "output_tokens": 45,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25183,
            "output_tokens": 51,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25266,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 26219,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26372,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 27248,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27340,
            "output_tokens": 64,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 27606,
            "output_tokens": 71,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28473,
            "output_tokens": 654,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29148,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 30230,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 31322,
            "output_tokens": 72,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31454,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31548,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31856,
            "output_tokens": 84,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32982,
            "output_tokens": 310,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33778,
            "output_tokens": 106,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34761,
            "output_tokens": 110,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35545,
            "output_tokens": 345,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36177,
            "output_tokens": 741,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36939,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37415,
            "output_tokens": 223,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37911,
            "output_tokens": 97,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 38394,
            "output_tokens": 97,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 39362,
            "output_tokens": 73,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 40482,
            "output_tokens": 70,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 41597,
            "output_tokens": 324,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42288,
            "output_tokens": 764,
            "reasoning_output_tokens": 556
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43339,
            "output_tokens": 440,
            "reasoning_output_tokens": 285
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 43934,
            "output_tokens": 347,
            "reasoning_output_tokens": 240
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44665,
            "output_tokens": 183,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45233,
            "output_tokens": 51,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45306,
            "output_tokens": 158,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45655,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 46112,
            "output_tokens": 102,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46596,
            "output_tokens": 98,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 46747,
            "output_tokens": 463,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 47233,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 60,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 17,
          "find": 1,
          "bash": 1,
          "devices": 4,
          "exec": 23,
          "device": 2,
          "write": 4,
          "device_stop": 4,
          "grep": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 400 rosbridge control frames; 70 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/phone-stand.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/workspace/.kinex/memos/phone-stand.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/workspace/.kinex/tools/cartesian_step.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-41-20261113/workspace/.kinex/tools/policy_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 114,
        "observed_images": 33
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 70,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 161,
        "published_frames": 161,
        "decode_ok": true,
        "duration_s": 4.025,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "8ca3fe7667ba55b5591c09713df15642c451afc0b2788283eefe96fb5766f50d",
        "source_sha256": {
          "third_person": "a2a65b339b6e7d0d73abfb1a8cf4e462617370fd15f441ef77dd1ec649ab507a",
          "left_wrist": "69bc290ed91d76ef4ceca031ada539719c4e04dccb1bdaa2aad31fd04f67220b",
          "right_wrist": "9d9f8e3b539e87141b22797228127a0c18305f139e0699b3ecc1e237b9cf7d36"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "cd389a2713d8",
          "task": "robotwin/place-phone-stand-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 162,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16000000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 162,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16000000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 162,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16000000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-42-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/42",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "42",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 17.76,
        "step_limit": 500,
        "steps": 444,
        "success": false,
        "valid": true
      },
      "steps": 444,
      "simulation_time_s": 17.76,
      "wall_time_s": 1434.364621,
      "timing": {
        "simulation_time_s": 17.76,
        "wall_time_s": 1434.364621,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the casual brown shoe off the table and drop it on the mat",
      "native_instruction": "Grab the casual brown shoe off the table and drop it on the mat",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9714524965060121,
        "cache_read_input_tokens": 1153152,
        "cache_reported_input_tokens": 1187039,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1187039,
        "cached_input_tokens": 1153152,
        "cost_usd": null,
        "fresh_input_tokens": 33887,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1187039,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1153152,
        "known_fresh_input_tokens": 33887,
        "known_input_tokens": 1187039,
        "known_output_tokens": 5639,
        "known_reasoning_output_tokens": 1192,
        "output_tokens": 5639,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1192,
        "reasoning_reported_output_tokens": 5639,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 40,
          "cache_write_input_tokens": 40,
          "cache_write_reported_input_tokens": 40,
          "cached_input_tokens": 40,
          "fresh_input_tokens": 40,
          "input_tokens": 40,
          "output_tokens": 40,
          "reasoning_output_tokens": 40,
          "reasoning_reported_output_tokens": 40
        },
        "request_attempts": 40,
        "response_count": 40,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1192678,
        "uncached_input_tokens": 33887,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11859,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11992,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14989,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 15278,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15104,
            "input_tokens": 15741,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 19698,
            "output_tokens": 168,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 22727,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23701,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24027,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24219,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24627,
            "output_tokens": 657,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25304,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 26175,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26264,
            "output_tokens": 242,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26532,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26736,
            "output_tokens": 122,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26884,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 29102,
            "output_tokens": 72,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 31358,
            "output_tokens": 102,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31486,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 32139,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32791,
            "output_tokens": 81,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33465,
            "output_tokens": 83,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 34138,
            "output_tokens": 81,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34815,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35469,
            "output_tokens": 116,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36176,
            "output_tokens": 76,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36462,
            "output_tokens": 241,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 36958,
            "output_tokens": 869,
            "reasoning_output_tokens": 329
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 37848,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38204,
            "output_tokens": 188,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38659,
            "output_tokens": 248,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39174,
            "output_tokens": 74,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39934,
            "output_tokens": 65,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 41066,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41425,
            "output_tokens": 359,
            "reasoning_output_tokens": 294
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42077,
            "output_tokens": 247,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42622,
            "output_tokens": 453,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43099,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 47,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 15,
          "find": 1,
          "bash": 1,
          "exec": 20,
          "devices": 1,
          "device": 1,
          "write": 3,
          "device_stop": 1,
          "edit": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 444 rosbridge control frames; 86 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-shoe.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/workspace/.kinex/memos/robotwin-shoe.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/workspace/.kinex/tools/cartesian_step.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/run_policy.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-42-20261113/workspace/.kinex/tools/run_policy.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 90,
        "observed_images": 34
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 86,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 179,
        "published_frames": 179,
        "decode_ok": true,
        "duration_s": 4.475,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "e7ab49870a517447f80aa8eb88c4f045256364db6c8449c14bfef6c546c4a61b",
        "source_sha256": {
          "third_person": "b5756701e7ac027ab7a042825cdef9391dc4b249819221b5e602c5fd901f5f9c",
          "left_wrist": "6b1d604583ae20de355e96dfb6c6e9b13b78fcb763be84b9a000c65de202ec5a",
          "right_wrist": "7fb49d8436d4ee2cedea6e740a01756918669bdd19991c5501791d2a7ce5ed14"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "db3c258fd3f1",
          "task": "robotwin/place-shoe-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 179,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 179,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17760000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 179,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 179,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17760000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 179,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 179,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17760000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-43-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/43",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "43",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 3.08,
        "step_limit": 400,
        "steps": 77,
        "success": true,
        "valid": true
      },
      "steps": 77,
      "simulation_time_s": 3.08,
      "wall_time_s": 511.73156,
      "timing": {
        "simulation_time_s": 3.08,
        "wall_time_s": 511.73156,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to press the shiny red plastic stapler.",
      "native_instruction": "Use the left arm to press the shiny red plastic stapler.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9696732167031646,
        "cache_read_input_tokens": 593792,
        "cache_reported_input_tokens": 612363,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 612363,
        "cached_input_tokens": 593792,
        "cost_usd": null,
        "fresh_input_tokens": 18571,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 612363,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 593792,
        "known_fresh_input_tokens": 18571,
        "known_input_tokens": 612363,
        "known_output_tokens": 4361,
        "known_reasoning_output_tokens": 504,
        "output_tokens": 4361,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 504,
        "reasoning_reported_output_tokens": 4361,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 29,
          "cache_write_input_tokens": 29,
          "cache_write_reported_input_tokens": 29,
          "cached_input_tokens": 29,
          "fresh_input_tokens": 29,
          "input_tokens": 29,
          "output_tokens": 29,
          "reasoning_output_tokens": 29,
          "reasoning_reported_output_tokens": 29
        },
        "request_attempts": 29,
        "response_count": 29,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 616724,
        "uncached_input_tokens": 18571,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11948,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11996,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12285,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 15888,
            "output_tokens": 102,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 17198,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17152,
            "input_tokens": 17768,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 17932,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18124,
            "output_tokens": 62,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18396,
            "output_tokens": 307,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 19098,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 19506,
            "output_tokens": 300,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20341,
            "output_tokens": 273,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 21768,
            "output_tokens": 356,
            "reasoning_output_tokens": 138
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22417,
            "output_tokens": 63,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23175,
            "output_tokens": 298,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23752,
            "output_tokens": 665,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24439,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24584,
            "output_tokens": 44,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24650,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25136,
            "output_tokens": 113,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25263,
            "output_tokens": 141,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25432,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25867,
            "output_tokens": 152,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26161,
            "output_tokens": 68,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26251,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26670,
            "output_tokens": 212,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27018,
            "output_tokens": 443,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27485,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 33,
        "model_tool_calls_by_name": {
          "read": 11,
          "ls": 1,
          "bash": 1,
          "exec": 10,
          "devices": 3,
          "device": 1,
          "write": 2,
          "device_stop": 3,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 77 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/press-stapler.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/workspace/.kinex/memos/press-stapler.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/left_cartesian.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-43-20261113/workspace/.kinex/tools/left_cartesian.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 65,
        "observed_images": 4
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 32,
        "published_frames": 32,
        "decode_ok": true,
        "duration_s": 0.8,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "73d8b08450930138796256fa526213f25a019deb1ec129dd045b8271ae01ed4e",
        "source_sha256": {
          "third_person": "9389b432ca7b03c5769e3a8f955c55911a4a8cccacd1c36515215a7445b5cb28",
          "left_wrist": "da46abd0daa904e3189743835461ee7f14f9474c6895188cd3c98d97b2f03cf0",
          "right_wrist": "0d0e3beee175c3b9a8852de4ce4b7d775b3b44488e9a2f46720dc7e4e5fc96d7"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "098a302387e4",
          "task": "robotwin/press-stapler-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 32,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 32,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3080000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 32,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 32,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3080000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 32,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 32,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3080000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-44-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/44",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "44",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.280000000000001,
        "step_limit": 400,
        "steps": 382,
        "success": false,
        "valid": true
      },
      "steps": 382,
      "simulation_time_s": 15.280000000000001,
      "wall_time_s": 1225.209256,
      "timing": {
        "simulation_time_s": 15.280000000000001,
        "wall_time_s": 1225.209256,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Catch the tabletop paymentsign, pick it up and rotate it, then set it back down.",
      "native_instruction": "Catch the tabletop paymentsign, pick it up and rotate it, then set it back down.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9717282815785038,
        "cache_read_input_tokens": 1527552,
        "cache_reported_input_tokens": 1571995,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1571995,
        "cached_input_tokens": 1527552,
        "cost_usd": null,
        "fresh_input_tokens": 44443,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1571995,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1527552,
        "known_fresh_input_tokens": 44443,
        "known_input_tokens": 1571995,
        "known_output_tokens": 6666,
        "known_reasoning_output_tokens": 951,
        "output_tokens": 6666,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 951,
        "reasoning_reported_output_tokens": 6666,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 52,
          "cache_write_input_tokens": 52,
          "cache_write_reported_input_tokens": 52,
          "cached_input_tokens": 52,
          "fresh_input_tokens": 52,
          "input_tokens": 52,
          "output_tokens": 52,
          "reasoning_output_tokens": 52,
          "reasoning_reported_output_tokens": 52
        },
        "request_attempts": 52,
        "response_count": 52,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1578661,
        "uncached_input_tokens": 44443,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11991,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12041,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13245,
            "output_tokens": 205,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13312,
            "input_tokens": 16860,
            "output_tokens": 220,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 20747,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 21927,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22253,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22444,
            "output_tokens": 70,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22883,
            "output_tokens": 706,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23612,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24496,
            "output_tokens": 39,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24597,
            "output_tokens": 242,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 24868,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25080,
            "output_tokens": 126,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25235,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26185,
            "output_tokens": 65,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 27317,
            "output_tokens": 92,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 28454,
            "output_tokens": 89,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28605,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28857,
            "output_tokens": 285,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29820,
            "output_tokens": 170,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30456,
            "output_tokens": 307,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31183,
            "output_tokens": 239,
            "reasoning_output_tokens": 197
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 32601,
            "output_tokens": 883,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33507,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33670,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34101,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34727,
            "output_tokens": 175,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35426,
            "output_tokens": 49,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35526,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35664,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35712,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35799,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36066,
            "output_tokens": 106,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36702,
            "output_tokens": 183,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36951,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37564,
            "output_tokens": 171,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38257,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38343,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38481,
            "output_tokens": 48,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38547,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38629,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38895,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39511,
            "output_tokens": 81,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39654,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39740,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39934,
            "output_tokens": 188,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40151,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40228,
            "output_tokens": 524,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 40777,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 62,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 16,
          "find": 1,
          "exec": 17,
          "devices": 3,
          "device": 9,
          "write": 3,
          "device_stop": 3,
          "edit": 4,
          "relative_pose": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 382 rosbridge control frames; 25 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/rotate-qrcode.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/workspace/.kinex/memos/rotate-qrcode.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-44-20261113/workspace/.kinex/tools/policy_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 117,
        "observed_images": 20
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 25,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 154,
        "published_frames": 154,
        "decode_ok": true,
        "duration_s": 3.85,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "0236acec51ec8012191bc59a8f9abfce5d34f3522a2c964b00c12d71e0f31e71",
        "source_sha256": {
          "third_person": "81f2d499677efedfbb484ae1ba78853a0ff3109bf957ab5a79af1cb8bd13d4d2",
          "left_wrist": "2330af8afb17463a062191e032f6b31623029c596f914c1d7c1beb6f01b7dcfa",
          "right_wrist": "4788e0ed930768504bdc0a1dc2fc38c003c22d7af731b139a98aeadcd3ee668c"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "464350faa6f7",
          "task": "robotwin/rotate-qrcode-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 154,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 154,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15280000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 154,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 154,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15280000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 154,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 154,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15280000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-45-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/45",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "45",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 17.52,
        "step_limit": 700,
        "steps": 438,
        "success": true,
        "valid": true
      },
      "steps": 438,
      "simulation_time_s": 17.52,
      "wall_time_s": 1364.872064,
      "timing": {
        "simulation_time_s": 17.52,
        "wall_time_s": 1364.872064,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the bottle with white cap with the left arm and shake.",
      "native_instruction": "Pick up the bottle with white cap with the left arm and shake.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9388219186413475,
        "cache_read_input_tokens": 1265792,
        "cache_reported_input_tokens": 1348277,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1348277,
        "cached_input_tokens": 1265792,
        "cost_usd": null,
        "fresh_input_tokens": 82485,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1348277,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1265792,
        "known_fresh_input_tokens": 82485,
        "known_input_tokens": 1348277,
        "known_output_tokens": 7180,
        "known_reasoning_output_tokens": 1130,
        "output_tokens": 7180,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1130,
        "reasoning_reported_output_tokens": 7180,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 41,
          "cache_write_input_tokens": 41,
          "cache_write_reported_input_tokens": 41,
          "cached_input_tokens": 41,
          "fresh_input_tokens": 41,
          "input_tokens": 41,
          "output_tokens": 41,
          "reasoning_output_tokens": 41,
          "reasoning_reported_output_tokens": 41
        },
        "request_attempts": 41,
        "response_count": 41,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1355457,
        "uncached_input_tokens": 82485,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11991,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14523,
            "output_tokens": 188,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14592,
            "input_tokens": 16864,
            "output_tokens": 143,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 19600,
            "output_tokens": 83,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20843,
            "output_tokens": 222,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 23506,
            "output_tokens": 155,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24785,
            "output_tokens": 83,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 25237,
            "output_tokens": 510,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25768,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25972,
            "output_tokens": 125,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 26124,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26973,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27058,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27476,
            "output_tokens": 227,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27748,
            "output_tokens": 187,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 27962,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 30824,
            "output_tokens": 86,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 34899,
            "output_tokens": 117,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35043,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 36542,
            "output_tokens": 46,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36650,
            "output_tokens": 71,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36931,
            "output_tokens": 330,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37868,
            "output_tokens": 95,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38429,
            "output_tokens": 149,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 39757,
            "output_tokens": 258,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40623,
            "output_tokens": 315,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41290,
            "output_tokens": 384,
            "reasoning_output_tokens": 96
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41995,
            "output_tokens": 844,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 42860,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43361,
            "output_tokens": 294,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 44105,
            "output_tokens": 219,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44722,
            "output_tokens": 267,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 45281,
            "output_tokens": 192,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45873,
            "output_tokens": 208,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45952,
            "input_tokens": 46357,
            "output_tokens": 225,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 46722,
            "output_tokens": 50,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 46834,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 47321,
            "output_tokens": 538,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 47884,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 61,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 16,
          "find": 1,
          "exec": 18,
          "device": 10,
          "write": 3,
          "edit": 4,
          "device_stop": 3,
          "devices": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 438 rosbridge control frames; 71 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/workspace/.kinex/tools/tcp_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vla_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-45-20261113/workspace/.kinex/tools/vla_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 105,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 71,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 176,
        "published_frames": 176,
        "decode_ok": true,
        "duration_s": 4.4,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "6d5b10060fd37ad2f7139f9a5e7d8d43058ab21d5a43a1c973725cc65aa317d4",
        "source_sha256": {
          "third_person": "c13c392d322e41afbd0c3543e1a6ccd0fbcca05e46c7928920611a9d1828f491",
          "left_wrist": "4bd24535d723d0b0533298b0c8f438f94a1c84578a3a15cd05a991cb53c57ae6",
          "right_wrist": "508aa104eca8fc654a39fd24c86aeba4ca2a28b297fe9b11b500c57c1b9826a7"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "c25edf43bc56",
          "task": "robotwin/shake-bottle-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 177,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 176,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17520000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 177,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 176,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17520000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 177,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 176,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17520000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-46-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/46",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "46",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 26.400000000000002,
        "step_limit": 700,
        "steps": 660,
        "success": false,
        "valid": true
      },
      "steps": 660,
      "simulation_time_s": 26.400000000000002,
      "wall_time_s": 1405.26802,
      "timing": {
        "simulation_time_s": 26.400000000000002,
        "wall_time_s": 1405.26802,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the bottle with white cap and shake it horizontally.",
      "native_instruction": "Grab the bottle with white cap and shake it horizontally.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.8911122345580622,
        "cache_read_input_tokens": 807296,
        "cache_reported_input_tokens": 905942,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 905942,
        "cached_input_tokens": 807296,
        "cost_usd": null,
        "fresh_input_tokens": 98646,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 905942,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 807296,
        "known_fresh_input_tokens": 98646,
        "known_input_tokens": 905942,
        "known_output_tokens": 3304,
        "known_reasoning_output_tokens": 538,
        "output_tokens": 3304,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 538,
        "reasoning_reported_output_tokens": 3304,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 32,
          "cache_write_input_tokens": 32,
          "cache_write_reported_input_tokens": 32,
          "cached_input_tokens": 32,
          "fresh_input_tokens": 32,
          "input_tokens": 32,
          "output_tokens": 32,
          "reasoning_output_tokens": 32,
          "reasoning_reported_output_tokens": 32
        },
        "request_attempts": 32,
        "response_count": 32,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 909246,
        "uncached_input_tokens": 98646,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11988,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12038,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 14766,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 20015,
            "output_tokens": 76,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 22851,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23421,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 23747,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24722,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24912,
            "output_tokens": 70,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25351,
            "output_tokens": 492,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25865,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26699,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26791,
            "output_tokens": 293,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27112,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27315,
            "output_tokens": 105,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 27448,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 29214,
            "output_tokens": 71,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 33231,
            "output_tokens": 106,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 33365,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 34511,
            "output_tokens": 66,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 35518,
            "output_tokens": 57,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35637,
            "output_tokens": 138,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36711,
            "output_tokens": 92,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 37732,
            "output_tokens": 72,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 38594,
            "output_tokens": 107,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39492,
            "output_tokens": 87,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39901,
            "output_tokens": 200,
            "reasoning_output_tokens": 141
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 40899,
            "output_tokens": 83,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41044,
            "output_tokens": 314,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41382,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 38,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 11,
          "find": 1,
          "exec": 13,
          "devices": 2,
          "device": 1,
          "write": 2,
          "device_stop": 3,
          "edit": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 660 rosbridge control frames; 165 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/workspace/.kinex/memos/robotwin.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/run_policy.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-46-20261113/workspace/.kinex/tools/run_policy.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 73,
        "observed_images": 12
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 165,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 265,
        "published_frames": 265,
        "decode_ok": true,
        "duration_s": 6.625,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "1a4968d03f25d863eb4e88b1f83f3b6dcb96ebde5a6b350d04736563503ffe0e",
        "source_sha256": {
          "third_person": "d9e4eac1bfb4bd0f4e56c2e06e19829f386dfd6096ee7f09fa3bf17d0c1e1826",
          "left_wrist": "62c71d5f63faa1f4447d6ea482025823389765689be32818ddeebbb1adab3652",
          "right_wrist": "d7a41a1cf9c61daa017197f579de6001050bb7f510c170d6afab7fdf083133e2"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "1da095ec867f",
          "task": "robotwin/shake-bottle-horizontally-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 266,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 265,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 26400000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 266,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 265,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 26400000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 266,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 265,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 26400000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-47-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/47",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "47",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 24.240000000000002,
        "step_limit": 800,
        "steps": 606,
        "success": true,
        "valid": true
      },
      "steps": 606,
      "simulation_time_s": 24.240000000000002,
      "wall_time_s": 1285.573839,
      "timing": {
        "simulation_time_s": 24.240000000000002,
        "wall_time_s": 1285.573839,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Bring red block and green block to the center, then position green block above red block.",
      "native_instruction": "Bring red block and green block to the center, then position green block above red block.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9574767568926105,
        "cache_read_input_tokens": 1478144,
        "cache_reported_input_tokens": 1543791,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1543791,
        "cached_input_tokens": 1478144,
        "cost_usd": null,
        "fresh_input_tokens": 65647,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1543791,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1478144,
        "known_fresh_input_tokens": 65647,
        "known_input_tokens": 1543791,
        "known_output_tokens": 7044,
        "known_reasoning_output_tokens": 872,
        "output_tokens": 7044,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 872,
        "reasoning_reported_output_tokens": 7044,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 50,
          "cache_write_input_tokens": 50,
          "cache_write_reported_input_tokens": 50,
          "cached_input_tokens": 50,
          "fresh_input_tokens": 50,
          "input_tokens": 50,
          "output_tokens": 50,
          "reasoning_output_tokens": 50,
          "reasoning_reported_output_tokens": 50
        },
        "request_attempts": 50,
        "response_count": 50,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1550835,
        "uncached_input_tokens": 65647,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11820,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11860,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11993,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12333,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 12559,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 16162,
            "output_tokens": 198,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16256,
            "input_tokens": 20003,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20329,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 21746,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 21938,
            "output_tokens": 52,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22359,
            "output_tokens": 560,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 22940,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23129,
            "output_tokens": 121,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23277,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 24134,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24222,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24785,
            "output_tokens": 85,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 26293,
            "output_tokens": 504,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12672,
            "input_tokens": 26824,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 28024,
            "output_tokens": 43,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 29413,
            "output_tokens": 76,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29699,
            "output_tokens": 386,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30489,
            "output_tokens": 269,
            "reasoning_output_tokens": 177
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 32401,
            "output_tokens": 380,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 32904,
            "output_tokens": 881,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33806,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34511,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34588,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34966,
            "output_tokens": 106,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35386,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35745,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36118,
            "output_tokens": 160,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36972,
            "output_tokens": 94,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37837,
            "output_tokens": 135,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38665,
            "output_tokens": 131,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 38931,
            "output_tokens": 90,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39156,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39534,
            "output_tokens": 67,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 39917,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 40307,
            "output_tokens": 71,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40691,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 41049,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41422,
            "output_tokens": 284,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41860,
            "output_tokens": 160,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42713,
            "output_tokens": 94,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43120,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 43311,
            "output_tokens": 42,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43415,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43725,
            "output_tokens": 662,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44410,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 59,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 18,
          "bash": 1,
          "exec": 27,
          "devices": 3,
          "device": 1,
          "write": 3,
          "edit": 3,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 606 rosbridge control frames; 49 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack-blocks.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/workspace/.kinex/memos/stack-blocks.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/run_vla.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-47-20261113/workspace/.kinex/tools/run_vla.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 112,
        "observed_images": 22
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 49,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 244,
        "published_frames": 244,
        "decode_ok": true,
        "duration_s": 6.1,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "aad207c9fe092bd2c58280ff56dec0a11e6353fbe1a4a7e7f4a32020fe36a0d9",
        "source_sha256": {
          "third_person": "1ae2f13d105ae668d278f634fbd65caa38b7cfe06314ff9687dfc3f79e63cdbb",
          "left_wrist": "15d4ac6cff9c812b8b0a845b3860991c8a0a7d868ad25e3df784407915d6c5c9",
          "right_wrist": "42777d7d1c0ab97277efef7ff1e0e10e416ad65e8e782f7a79adb607167cf13a"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "014b891ed90a",
          "task": "robotwin/stack-blocks-two-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 244,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 244,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24240000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 244,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 244,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24240000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 244,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 244,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24240000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-48-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/48",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "48",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 28.560000000000002,
        "step_limit": 900,
        "steps": 714,
        "success": true,
        "valid": true
      },
      "steps": 714,
      "simulation_time_s": 28.560000000000002,
      "wall_time_s": 1497.46872,
      "timing": {
        "simulation_time_s": 28.560000000000002,
        "wall_time_s": 1497.46872,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Stack the small smooth brown-rimmed bowl directly over the smooth bowl with glossy finish.",
      "native_instruction": "Stack the small smooth brown-rimmed bowl directly over the smooth bowl with glossy finish.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9751715725612301,
        "cache_read_input_tokens": 1518976,
        "cache_reported_input_tokens": 1557650,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1557650,
        "cached_input_tokens": 1518976,
        "cost_usd": null,
        "fresh_input_tokens": 38674,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1557650,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1518976,
        "known_fresh_input_tokens": 38674,
        "known_input_tokens": 1557650,
        "known_output_tokens": 9056,
        "known_reasoning_output_tokens": 2236,
        "output_tokens": 9056,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2236,
        "reasoning_reported_output_tokens": 9056,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 46,
          "cache_write_input_tokens": 46,
          "cache_write_reported_input_tokens": 46,
          "cached_input_tokens": 46,
          "fresh_input_tokens": 46,
          "input_tokens": 46,
          "output_tokens": 46,
          "reasoning_output_tokens": 46,
          "reasoning_reported_output_tokens": 46
        },
        "request_attempts": 46,
        "response_count": 46,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1566706,
        "uncached_input_tokens": 38674,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11859,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11992,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 15543,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15488,
            "input_tokens": 15832,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 19435,
            "output_tokens": 230,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 23844,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24170,
            "output_tokens": 132,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25746,
            "output_tokens": 680,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26449,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26658,
            "output_tokens": 118,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26803,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27634,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28461,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 29288,
            "output_tokens": 82,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 30137,
            "output_tokens": 57,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30256,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30508,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 31012,
            "output_tokens": 378,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 32278,
            "output_tokens": 632,
            "reasoning_output_tokens": 320
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33164,
            "output_tokens": 291,
            "reasoning_output_tokens": 249
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 34634,
            "output_tokens": 862,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35517,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35973,
            "output_tokens": 151,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36497,
            "output_tokens": 124,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36993,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37406,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37861,
            "output_tokens": 662,
            "reasoning_output_tokens": 272
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38991,
            "output_tokens": 426,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39702,
            "output_tokens": 175,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40566,
            "output_tokens": 413,
            "reasoning_output_tokens": 281
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41349,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41769,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42218,
            "output_tokens": 124,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42716,
            "output_tokens": 103,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43189,
            "output_tokens": 122,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43683,
            "output_tokens": 189,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 44241,
            "output_tokens": 611,
            "reasoning_output_tokens": 186
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45131,
            "output_tokens": 376,
            "reasoning_output_tokens": 219
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45875,
            "output_tokens": 90,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46111,
            "output_tokens": 65,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45952,
            "input_tokens": 46238,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46431,
            "output_tokens": 145,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46957,
            "output_tokens": 60,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47044,
            "output_tokens": 601,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47488,
            "input_tokens": 47670,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 57,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 16,
          "find": 1,
          "bash": 1,
          "exec": 27,
          "devices": 2,
          "device": 2,
          "write": 3,
          "edit": 2,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 714 rosbridge control frames; 60 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-bowls.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/workspace/.kinex/memos/robotwin-bowls.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-48-20261113/workspace/.kinex/tools/policy_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 106,
        "observed_images": 29
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 60,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 287,
        "published_frames": 287,
        "decode_ok": true,
        "duration_s": 7.175,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "7293424757518ddae582be053c23132f380dc0ddc67f8c4da11dd6c9b38bffe4",
        "source_sha256": {
          "third_person": "48025ad2d8e5ff2ee6cdfa519c08b7037c1358c027e315ebbf14c003ba2ffa98",
          "left_wrist": "782b3d8723ec1e7d945ed20bb6ea416fb090c79f742d8657134c28e3fb57733e",
          "right_wrist": "569d3c44aa1a445148fc9c836bb6b9b3cd19e546f9871e94b1184daf262b006a"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ea74f5805220",
          "task": "robotwin/stack-bowls-two-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 287,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 287,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 287,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 287,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 287,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 287,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-49-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/49",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "49",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.96,
        "step_limit": 400,
        "steps": 399,
        "success": false,
        "valid": true
      },
      "steps": 399,
      "simulation_time_s": 15.96,
      "wall_time_s": 1078.498606,
      "timing": {
        "simulation_time_s": 15.96,
        "wall_time_s": 1078.498606,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Stamp Green using the block-shaped seal with angled edges",
      "native_instruction": "Stamp Green using the block-shaped seal with angled edges",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9651856899377574,
        "cache_read_input_tokens": 1181312,
        "cache_reported_input_tokens": 1223922,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1223922,
        "cached_input_tokens": 1181312,
        "cost_usd": null,
        "fresh_input_tokens": 42610,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1223922,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1181312,
        "known_fresh_input_tokens": 42610,
        "known_input_tokens": 1223922,
        "known_output_tokens": 5042,
        "known_reasoning_output_tokens": 469,
        "output_tokens": 5042,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 469,
        "reasoning_reported_output_tokens": 5042,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 40,
          "cache_write_input_tokens": 40,
          "cache_write_reported_input_tokens": 40,
          "cached_input_tokens": 40,
          "fresh_input_tokens": 40,
          "input_tokens": 40,
          "output_tokens": 40,
          "reasoning_output_tokens": 40,
          "reasoning_reported_output_tokens": 40
        },
        "request_attempts": 40,
        "response_count": 40,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1228964,
        "uncached_input_tokens": 42610,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11814,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11854,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11987,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14416,
            "output_tokens": 217,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 17322,
            "output_tokens": 202,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 21924,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22250,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 23667,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 23858,
            "output_tokens": 57,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24284,
            "output_tokens": 513,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24818,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25021,
            "output_tokens": 39,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25122,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25246,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26080,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26169,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26587,
            "output_tokens": 71,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 28697,
            "output_tokens": 608,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29326,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 31709,
            "output_tokens": 77,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 34804,
            "output_tokens": 105,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 34936,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35362,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35794,
            "output_tokens": 82,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36257,
            "output_tokens": 64,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36531,
            "output_tokens": 248,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37109,
            "output_tokens": 87,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37366,
            "output_tokens": 679,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38066,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38456,
            "output_tokens": 76,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38870,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39267,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39671,
            "output_tokens": 120,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40130,
            "output_tokens": 100,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40570,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40973,
            "output_tokens": 83,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41395,
            "output_tokens": 113,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41845,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 41937,
            "output_tokens": 471,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42432,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 51,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 15,
          "find": 1,
          "devices": 2,
          "exec": 18,
          "device": 2,
          "write": 4,
          "device_stop": 3,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 399 rosbridge control frames; 68 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-stamping.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/workspace/.kinex/memos/robotwin-stamping.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-49-20261113/workspace/.kinex/tools/policy_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 94,
        "observed_images": 17
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 68,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 161,
        "published_frames": 161,
        "decode_ok": true,
        "duration_s": 4.025,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "8e57a38db4ea6dfb6ba8997534b0af028103ce8568ae3c08e1f8d85ebf9cf98b",
        "source_sha256": {
          "third_person": "8c763aed119b1df6c13a0b882ca339855ddec380ad9ca270e12f89158f0372b0",
          "left_wrist": "1555ad9e41e6ec575b7912e4488115e968b4c7c2f987f4cc62e27990391a8754",
          "right_wrist": "ec45285cf790ecbec01b92aa3fc797223261ad4fbcf7516c18e438f6db5cce73"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "376b8ef89fc6",
          "task": "robotwin/stamp-seal-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "native-kinex-vlact50k-50-20261113",
      "variant": "kinex-vlact50k-native",
      "task_key": "task03/50",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "50",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 13.64,
        "step_limit": 400,
        "steps": 341,
        "success": true,
        "valid": true
      },
      "steps": 341,
      "simulation_time_s": 13.64,
      "wall_time_s": 1056.638316,
      "timing": {
        "simulation_time_s": 13.64,
        "wall_time_s": 1056.638316,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Click the flat gray switch with the right arm",
      "native_instruction": "Click the flat gray switch with the right arm",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9647657316682321,
        "cache_read_input_tokens": 1041536,
        "cache_reported_input_tokens": 1079574,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1079574,
        "cached_input_tokens": 1041536,
        "cost_usd": null,
        "fresh_input_tokens": 38038,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1079574,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1041536,
        "known_fresh_input_tokens": 38038,
        "known_input_tokens": 1079574,
        "known_output_tokens": 5446,
        "known_reasoning_output_tokens": 1179,
        "output_tokens": 5446,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1179,
        "reasoning_reported_output_tokens": 5446,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 38,
          "cache_write_input_tokens": 38,
          "cache_write_reported_input_tokens": 38,
          "cached_input_tokens": 38,
          "fresh_input_tokens": 38,
          "input_tokens": 38,
          "output_tokens": 38,
          "reasoning_output_tokens": 38,
          "reasoning_reported_output_tokens": 38
        },
        "request_attempts": 38,
        "response_count": 38,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1085020,
        "uncached_input_tokens": 38038,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11812,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11945,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11995,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 12284,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 13101,
            "output_tokens": 162,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13184,
            "input_tokens": 15898,
            "output_tokens": 103,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 17177,
            "output_tokens": 175,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17152,
            "input_tokens": 20988,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21314,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21505,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 22479,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22887,
            "output_tokens": 533,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23441,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23642,
            "output_tokens": 126,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 23795,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24807,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 27327,
            "output_tokens": 79,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 29106,
            "output_tokens": 75,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 30891,
            "output_tokens": 305,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31786,
            "output_tokens": 188,
            "reasoning_output_tokens": 144
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32184,
            "output_tokens": 263,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33204,
            "output_tokens": 362,
            "reasoning_output_tokens": 327
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33628,
            "output_tokens": 696,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34345,
            "output_tokens": 103,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34795,
            "output_tokens": 92,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35572,
            "output_tokens": 275,
            "reasoning_output_tokens": 192
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36194,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36622,
            "output_tokens": 96,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37062,
            "output_tokens": 47,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37279,
            "output_tokens": 104,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37405,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37826,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38016,
            "output_tokens": 103,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38585,
            "output_tokens": 151,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39085,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39310,
            "output_tokens": 79,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39879,
            "output_tokens": 502,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40403,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 45,
        "model_tool_calls_by_name": {
          "read": 15,
          "ls": 1,
          "bash": 1,
          "exec": 17,
          "devices": 2,
          "device": 3,
          "write": 3,
          "edit": 1,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 341 rosbridge control frames; 40 optional VLA calls.",
        "instruction_alignment": "Verbatim native RoboTwin episode instruction.",
        "confidence": "One authored seed per task in the full 50-task Hard suite."
      },
      "links": {
        "video": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/transcript.json",
        "usage": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/usage.json",
        "analysis": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/analysis.json",
        "protocol": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/protocol.json",
        "native_goal": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/instruction.json",
        "verdict": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/episode.json",
        "media_check": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/evidence.zip",
        "timing": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/timing.json",
        "native_session": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/session.jsonl",
        "trace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/trace.json",
        "provider_audit": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/turn-switch.md",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/workspace/.kinex/memos/turn-switch.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/move_tcp.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/workspace/.kinex/tools/move_tcp.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/policy_step.py",
          "file": "experiments/kinex-hard-native-20261113/native-kinex-vlact50k-50-20261113/workspace/.kinex/tools/policy_step.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 86,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "native-model-device-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-and-task-instruction",
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 40,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 138,
        "published_frames": 138,
        "decode_ok": true,
        "duration_s": 3.45,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "56e00ebd5fb18c9ae922f9f7689b689ed7328c976d0c508fb6cef84fcb5b8369",
        "source_sha256": {
          "third_person": "d9855b68ecc6b30597ba840df348a2f8aa6ef8711adb05295a54467387ed5cce",
          "left_wrist": "fdb1c5a5026198ae8b3970cb86ba14e918045e1c51ce79b928286cbbb54a196c",
          "right_wrist": "c6edb614a2b25dfb5a94c60c43692ced41cea0f66ead53f765ce5d5bc407f989"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "8333b39c9969",
          "task": "robotwin/turn-switch-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 138,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 138,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 138,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "native-model-device-20261113"
    },
    {
      "id": "onboard-rerun-kinex-tcp-posestate-01-20261115",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/01",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "01",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 24.92,
        "step_limit": 800,
        "steps": 623,
        "success": true,
        "valid": true
      },
      "steps": 623,
      "simulation_time_s": 24.92,
      "wall_time_s": 1150.646611,
      "timing": {
        "simulation_time_s": 24.92,
        "wall_time_s": 1150.646611,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "native_instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9719748375978966,
        "cache_read_input_tokens": 1456896,
        "cache_reported_input_tokens": 1498903,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1498903,
        "cached_input_tokens": 1456896,
        "cost_usd": null,
        "fresh_input_tokens": 42007,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1498903,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1456896,
        "known_fresh_input_tokens": 42007,
        "known_input_tokens": 1498903,
        "known_output_tokens": 9532,
        "known_reasoning_output_tokens": 2592,
        "output_tokens": 9532,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2592,
        "reasoning_reported_output_tokens": 9532,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 46,
          "cache_write_input_tokens": 46,
          "cache_write_reported_input_tokens": 46,
          "cached_input_tokens": 46,
          "fresh_input_tokens": 46,
          "input_tokens": 46,
          "output_tokens": 46,
          "reasoning_output_tokens": 46,
          "reasoning_reported_output_tokens": 46
        },
        "request_attempts": 46,
        "response_count": 46,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1508435,
        "uncached_input_tokens": 42007,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11826,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11866,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11999,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14351,
            "output_tokens": 245,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 18708,
            "output_tokens": 181,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 22507,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22835,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 24234,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25313,
            "output_tokens": 136,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25844,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26252,
            "output_tokens": 139,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26702,
            "output_tokens": 344,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27300,
            "output_tokens": 391,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27993,
            "output_tokens": 1140,
            "reasoning_output_tokens": 298
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29157,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29927,
            "output_tokens": 49,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30038,
            "output_tokens": 62,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30310,
            "output_tokens": 68,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31557,
            "output_tokens": 928,
            "reasoning_output_tokens": 120
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32508,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32935,
            "output_tokens": 173,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33456,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33886,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34298,
            "output_tokens": 238,
            "reasoning_output_tokens": 154
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34880,
            "output_tokens": 56,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35258,
            "output_tokens": 197,
            "reasoning_output_tokens": 116
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35804,
            "output_tokens": 122,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36273,
            "output_tokens": 89,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36424,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36786,
            "output_tokens": 384,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37618,
            "output_tokens": 293,
            "reasoning_output_tokens": 210
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38259,
            "output_tokens": 350,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38868,
            "output_tokens": 154,
            "reasoning_output_tokens": 113
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39488,
            "output_tokens": 156,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39992,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40403,
            "output_tokens": 105,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40858,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41291,
            "output_tokens": 209,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41848,
            "output_tokens": 103,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42298,
            "output_tokens": 65,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42425,
            "output_tokens": 492,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43697,
            "output_tokens": 828,
            "reasoning_output_tokens": 731
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44667,
            "output_tokens": 41,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 44878,
            "output_tokens": 296,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45204,
            "output_tokens": 653,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 45882,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 56,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 17,
          "find": 1,
          "devices": 3,
          "exec": 24,
          "write": 3,
          "device_stop": 3,
          "device": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 623 rosbridge control frames; 40 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; onboard rerun of the first five Hard tasks.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 15
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-rerun-20261115/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/handover-block.md",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/workspace/.kinex/memos/handover-block.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-01-20261115/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 105,
        "observed_images": 25
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 40,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 250,
        "published_frames": 250,
        "decode_ok": true,
        "duration_s": 6.25,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "02ceb4552af36913546f2fa283268e2d9e222f1f3c497e8acac450a66e283055",
        "source_sha256": {
          "third_person": "38947d82c7e960a887dcf2d45810fc08b0c08a23c3fb46c818689a72c0218948",
          "left_wrist": "f91f5b6f15e95ae3d364a314cf5543f5dadc1b060e2480c3a15dc750f1f5dc36",
          "right_wrist": "38802fe6561e3713017d1de1221d2a51c272ea274d5b53fcbcd7c85c5c2facbe"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "4bfbad21f1d6",
          "task": "robotwin/handover-block-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 251,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 250,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24920000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 251,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 250,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24920000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 251,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 250,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24920000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard5-20261115",
      "evaluation_date": "2026-11-15",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-rerun-kinex-tcp-posestate-02-20261115",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/02",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "02",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 23.8,
        "step_limit": 600,
        "steps": 595,
        "success": false,
        "valid": true
      },
      "steps": 595,
      "simulation_time_s": 23.8,
      "wall_time_s": 1309.522277,
      "timing": {
        "simulation_time_s": 23.8,
        "wall_time_s": 1309.522277,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "native_instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9653412500018611,
        "cache_read_input_tokens": 1296768,
        "cache_reported_input_tokens": 1343326,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1343326,
        "cached_input_tokens": 1296768,
        "cost_usd": null,
        "fresh_input_tokens": 46558,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1343326,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1296768,
        "known_fresh_input_tokens": 46558,
        "known_input_tokens": 1343326,
        "known_output_tokens": 8582,
        "known_reasoning_output_tokens": 2553,
        "output_tokens": 8582,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2553,
        "reasoning_reported_output_tokens": 8582,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 39,
          "cache_write_input_tokens": 39,
          "cache_write_reported_input_tokens": 39,
          "cached_input_tokens": 39,
          "fresh_input_tokens": 39,
          "input_tokens": 39,
          "output_tokens": 39,
          "reasoning_output_tokens": 39,
          "reasoning_reported_output_tokens": 39
        },
        "request_attempts": 39,
        "response_count": 39,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1351908,
        "uncached_input_tokens": 46558,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11991,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12041,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12334,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18224,
            "output_tokens": 158,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 23634,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 23962,
            "output_tokens": 169,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 27037,
            "output_tokens": 583,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28378,
            "output_tokens": 1117,
            "reasoning_output_tokens": 239
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29518,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29752,
            "output_tokens": 116,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 31099,
            "output_tokens": 145,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 32476,
            "output_tokens": 161,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33356,
            "output_tokens": 67,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 34412,
            "output_tokens": 800,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35233,
            "output_tokens": 119,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35843,
            "output_tokens": 83,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36302,
            "output_tokens": 119,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36908,
            "output_tokens": 423,
            "reasoning_output_tokens": 207
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37742,
            "output_tokens": 311,
            "reasoning_output_tokens": 226
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38481,
            "output_tokens": 153,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 39068,
            "output_tokens": 47,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39177,
            "output_tokens": 477,
            "reasoning_output_tokens": 164
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40030,
            "output_tokens": 242,
            "reasoning_output_tokens": 174
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40647,
            "output_tokens": 105,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 41437,
            "output_tokens": 60,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 42456,
            "output_tokens": 104,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42988,
            "output_tokens": 288,
            "reasoning_output_tokens": 165
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43773,
            "output_tokens": 229,
            "reasoning_output_tokens": 164
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44377,
            "output_tokens": 135,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44949,
            "output_tokens": 178,
            "reasoning_output_tokens": 92
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45558,
            "output_tokens": 256,
            "reasoning_output_tokens": 173
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 46242,
            "output_tokens": 234,
            "reasoning_output_tokens": 144
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46908,
            "output_tokens": 136,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47424,
            "output_tokens": 106,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47360,
            "input_tokens": 47963,
            "output_tokens": 243,
            "reasoning_output_tokens": 154
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48000,
            "input_tokens": 48643,
            "output_tokens": 621,
            "reasoning_output_tokens": 157
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49152,
            "input_tokens": 49287,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 53,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 19,
          "bash": 1,
          "exec": 25,
          "devices": 1,
          "device": 1,
          "write": 3,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 595 rosbridge control frames; 33 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; onboard rerun of the first five Hard tasks.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 17
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-rerun-20261115/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/dual_shoes.md",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/workspace/.kinex/memos/dual_shoes.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-02-20261115/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 95,
        "observed_images": 25
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 33,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 239,
        "published_frames": 239,
        "decode_ok": true,
        "duration_s": 5.975,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "a1d288271c02777312b7c73a0c7292d11a1c968e382a28bbb4b7ac739acccdae",
        "source_sha256": {
          "third_person": "f1a22a7d8f482f7232bab5d552db4ad7b057576f75e2cda68734df752a55e766",
          "left_wrist": "1dbd1d7c7b52b351f8df91bd33a8eb977a5386fb42cd33753964f32118de345b",
          "right_wrist": "7ad46ac8e61f74e81a1da0b3c260a0d7bc85cd7e1088de07230be64355f1b27b"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "9bc2eb5373ba",
          "task": "robotwin/place-dual-shoes-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 240,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 239,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23800000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 240,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 239,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23800000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 240,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 239,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23800000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard5-20261115",
      "evaluation_date": "2026-11-15",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-rerun-kinex-tcp-posestate-03-20261115",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/03",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "03",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 2
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 47.64,
        "step_limit": 1700,
        "steps": 1191,
        "success": true,
        "valid": true
      },
      "steps": 1191,
      "simulation_time_s": 47.64,
      "wall_time_s": 1670.475165,
      "timing": {
        "simulation_time_s": 47.64,
        "wall_time_s": 1670.475165,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap in the dustbin the simple black garbage bin.",
      "native_instruction": "Place the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap in the dustbin the simple black garbage bin.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9772599295819415,
        "cache_read_input_tokens": 2382848,
        "cache_reported_input_tokens": 2438295,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2438295,
        "cached_input_tokens": 2382848,
        "cost_usd": null,
        "fresh_input_tokens": 55447,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2438295,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2382848,
        "known_fresh_input_tokens": 55447,
        "known_input_tokens": 2438295,
        "known_output_tokens": 10216,
        "known_reasoning_output_tokens": 2479,
        "output_tokens": 10216,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2479,
        "reasoning_reported_output_tokens": 10216,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 58,
          "cache_write_input_tokens": 58,
          "cache_write_reported_input_tokens": 58,
          "cached_input_tokens": 58,
          "fresh_input_tokens": 58,
          "input_tokens": 58,
          "output_tokens": 58,
          "reasoning_output_tokens": 58,
          "reasoning_reported_output_tokens": 58
        },
        "request_attempts": 58,
        "response_count": 58,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2448511,
        "uncached_input_tokens": 55447,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11843,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11883,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12016,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15490,
            "output_tokens": 245,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15616,
            "input_tokens": 19822,
            "output_tokens": 225,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 26873,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27201,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27393,
            "output_tokens": 213,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 30664,
            "output_tokens": 509,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31195,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 32050,
            "output_tokens": 262,
            "reasoning_output_tokens": 220
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32778,
            "output_tokens": 915,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33717,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34525,
            "output_tokens": 130,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35460,
            "output_tokens": 68,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35738,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36767,
            "output_tokens": 945,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37735,
            "output_tokens": 113,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38542,
            "output_tokens": 96,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39333,
            "output_tokens": 230,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39872,
            "output_tokens": 154,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40248,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40622,
            "output_tokens": 220,
            "reasoning_output_tokens": 106
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41739,
            "output_tokens": 328,
            "reasoning_output_tokens": 246
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42385,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42577,
            "output_tokens": 245,
            "reasoning_output_tokens": 164
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 43138,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 43331,
            "output_tokens": 113,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 44140,
            "output_tokens": 58,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 45119,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45245,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45636,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 45829,
            "output_tokens": 214,
            "reasoning_output_tokens": 131
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46778,
            "output_tokens": 90,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 47182,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47375,
            "output_tokens": 191,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47488,
            "input_tokens": 48216,
            "output_tokens": 257,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 48683,
            "output_tokens": 474,
            "reasoning_output_tokens": 339
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49480,
            "output_tokens": 150,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 49850,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49792,
            "input_tokens": 50227,
            "output_tokens": 134,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50176,
            "input_tokens": 50680,
            "output_tokens": 144,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50688,
            "input_tokens": 51031,
            "output_tokens": 172,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51072,
            "input_tokens": 51517,
            "output_tokens": 151,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51584,
            "input_tokens": 51930,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51840,
            "input_tokens": 52303,
            "output_tokens": 284,
            "reasoning_output_tokens": 173
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52480,
            "input_tokens": 52904,
            "output_tokens": 131,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52864,
            "input_tokens": 53242,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53120,
            "input_tokens": 53634,
            "output_tokens": 68,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53504,
            "input_tokens": 54183,
            "output_tokens": 167,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54144,
            "input_tokens": 54670,
            "output_tokens": 152,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54656,
            "input_tokens": 55044,
            "output_tokens": 65,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55040,
            "input_tokens": 55428,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55424,
            "input_tokens": 55822,
            "output_tokens": 118,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55808,
            "input_tokens": 56081,
            "output_tokens": 95,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56064,
            "input_tokens": 56556,
            "output_tokens": 317,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56704,
            "input_tokens": 56903,
            "output_tokens": 740,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57472,
            "input_tokens": 57670,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 72,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 19,
          "find": 1,
          "exec": 35,
          "devices": 3,
          "device": 5,
          "write": 4,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 1191 rosbridge control frames; 65 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; onboard rerun of the first five Hard tasks.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 27
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-rerun-20261115/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/bottles-dustbin.md",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/workspace/.kinex/memos/bottles-dustbin.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/observe.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/workspace/.kinex/tools/observe.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-03-20261115/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 133,
        "observed_images": 40
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 1700,
        "max_control_steps": 1700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 65,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 478,
        "published_frames": 478,
        "decode_ok": true,
        "duration_s": 11.95,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f935957a70b67f09d769738f313377fef45a40cece6a752f5b6b347ebe2ccbd6",
        "source_sha256": {
          "third_person": "cdc10fd6a60c1dff209c82cafbe31c033df37e1224b2635ff6709c20f61592d7",
          "left_wrist": "eba8652740bbd7047cba923528c90dccf880f2af004846f889088e723833398a",
          "right_wrist": "a2ac00499f12fb7b2041fa63db41badcbc0a6a62bcb1527ac6a58f22dd730a06"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "81022333656b",
          "task": "robotwin/put-bottles-dustbin-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 478,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 478,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 47640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 478,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 478,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 47640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 478,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 478,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 47640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard5-20261115",
      "evaluation_date": "2026-11-15",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-rerun-kinex-tcp-posestate-04-20261115",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/04",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "04",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 14.96,
        "step_limit": 500,
        "steps": 374,
        "success": false,
        "valid": true
      },
      "steps": 374,
      "simulation_time_s": 14.96,
      "wall_time_s": 944.771242,
      "timing": {
        "simulation_time_s": 14.96,
        "wall_time_s": 944.771242,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "native_instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.961188285847754,
        "cache_read_input_tokens": 932096,
        "cache_reported_input_tokens": 969733,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 969733,
        "cached_input_tokens": 932096,
        "cost_usd": null,
        "fresh_input_tokens": 37637,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 969733,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 932096,
        "known_fresh_input_tokens": 37637,
        "known_input_tokens": 969733,
        "known_output_tokens": 8653,
        "known_reasoning_output_tokens": 3654,
        "output_tokens": 8653,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3654,
        "reasoning_reported_output_tokens": 8653,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 37,
          "cache_write_input_tokens": 37,
          "cache_write_reported_input_tokens": 37,
          "cached_input_tokens": 37,
          "fresh_input_tokens": 37,
          "input_tokens": 37,
          "output_tokens": 37,
          "reasoning_output_tokens": 37,
          "reasoning_reported_output_tokens": 37
        },
        "request_attempts": 37,
        "response_count": 37,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 978386,
        "uncached_input_tokens": 37637,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11827,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11867,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12000,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12344,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 12807,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12672,
            "input_tokens": 14032,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13952,
            "input_tokens": 14846,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14720,
            "input_tokens": 15663,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15616,
            "input_tokens": 16489,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 17317,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17152,
            "input_tokens": 18109,
            "output_tokens": 140,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 20437,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 22043,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22371,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22563,
            "output_tokens": 124,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 24093,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24753,
            "output_tokens": 145,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25364,
            "output_tokens": 935,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26321,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 27638,
            "output_tokens": 251,
            "reasoning_output_tokens": 139
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 29092,
            "output_tokens": 109,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 30398,
            "output_tokens": 981,
            "reasoning_output_tokens": 189
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31400,
            "output_tokens": 110,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31907,
            "output_tokens": 71,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32356,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32837,
            "output_tokens": 278,
            "reasoning_output_tokens": 239
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33438,
            "output_tokens": 150,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33912,
            "output_tokens": 831,
            "reasoning_output_tokens": 716
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35129,
            "output_tokens": 174,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35510,
            "output_tokens": 327,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36114,
            "output_tokens": 726,
            "reasoning_output_tokens": 410
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37174,
            "output_tokens": 540,
            "reasoning_output_tokens": 477
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38101,
            "output_tokens": 141,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38620,
            "output_tokens": 609,
            "reasoning_output_tokens": 545
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39617,
            "output_tokens": 220,
            "reasoning_output_tokens": 170
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40213,
            "output_tokens": 796,
            "reasoning_output_tokens": 194
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41031,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 41,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 17,
          "bash": 1,
          "exec": 16,
          "devices": 2,
          "device": 1,
          "write": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 374 rosbridge control frames; 36 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; onboard rerun of the first five Hard tasks.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 8
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-rerun-20261115/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/scan-object.md",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/workspace/.kinex/memos/scan-object.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-04-20261115/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 81,
        "observed_images": 19
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 36,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 151,
        "published_frames": 151,
        "decode_ok": true,
        "duration_s": 3.775,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "8a8af8cc2efba4bee0714ff8acf4cc882a6467ec9751d4119fbcbe3346333025",
        "source_sha256": {
          "third_person": "2f15ec8433182f7a389d8ab3bdfcf273924c8e1356c435f2b33019ecbe7f5f5f",
          "left_wrist": "08300d55eef8b7b7b4ebd5d7a5515a662da34f19f80754f549b3e1ba024737e4",
          "right_wrist": "facdd52691bfffd708802cba624a5dbe4ed45c9b293af6819a35d5f946d9f099"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ea2fa261e896",
          "task": "robotwin/scan-object-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 151,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 151,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 151,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 151,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 151,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 151,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard5-20261115",
      "evaluation_date": "2026-11-15",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-rerun-kinex-tcp-posestate-05-20261115",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/05",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "05",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 11.32,
        "step_limit": 1200,
        "steps": 283,
        "success": false,
        "valid": true
      },
      "steps": 283,
      "simulation_time_s": 11.32,
      "wall_time_s": 668.825489,
      "timing": {
        "simulation_time_s": 11.32,
        "wall_time_s": 668.825489,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "native_instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9553307671649808,
        "cache_read_input_tokens": 683008,
        "cache_reported_input_tokens": 714944,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 714944,
        "cached_input_tokens": 683008,
        "cost_usd": null,
        "fresh_input_tokens": 31936,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 714944,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 683008,
        "known_fresh_input_tokens": 31936,
        "known_input_tokens": 714944,
        "known_output_tokens": 5518,
        "known_reasoning_output_tokens": 903,
        "output_tokens": 5518,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 903,
        "reasoning_reported_output_tokens": 5518,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 29,
          "cache_write_input_tokens": 29,
          "cache_write_reported_input_tokens": 29,
          "cached_input_tokens": 29,
          "fresh_input_tokens": 29,
          "input_tokens": 29,
          "output_tokens": 29,
          "reasoning_output_tokens": 29,
          "reasoning_reported_output_tokens": 29
        },
        "request_attempts": 29,
        "response_count": 29,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 720462,
        "uncached_input_tokens": 31936,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11819,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11859,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11992,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12042,
            "output_tokens": 145,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12401,
            "output_tokens": 168,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16513,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 20329,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 21728,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 22056,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22245,
            "output_tokens": 91,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 23375,
            "output_tokens": 316,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24086,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24494,
            "output_tokens": 316,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25229,
            "output_tokens": 301,
            "reasoning_output_tokens": 261
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25996,
            "output_tokens": 1014,
            "reasoning_output_tokens": 127
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27032,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27838,
            "output_tokens": 95,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28143,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 29172,
            "output_tokens": 917,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30110,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30529,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30994,
            "output_tokens": 78,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31242,
            "output_tokens": 51,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31355,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31718,
            "output_tokens": 364,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32272,
            "output_tokens": 100,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32423,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32615,
            "output_tokens": 698,
            "reasoning_output_tokens": 110
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33337,
            "output_tokens": 42,
            "reasoning_output_tokens": 9
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 40,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 17,
          "find": 4,
          "exec": 7,
          "devices": 2,
          "device": 4,
          "write": 3,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 283 rosbridge control frames; 44 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; onboard rerun of the first five Hard tasks.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 2
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs.",
        "observed_anomaly": "All three bowls, including the two untouched ones, left every camera view during the agent's direct-control descent onto the rear bowl, while other objects stayed in place (see poster and terminal frames). The native verifier marked the episode valid; the cause was not diagnosed and the outcome was not retried."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-rerun-20261115/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack_bowls.md",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/workspace/.kinex/memos/stack_bowls.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-20261115/onboard-rerun-kinex-tcp-posestate-05-20261115/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 72,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 44,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 114,
        "published_frames": 114,
        "decode_ok": true,
        "duration_s": 2.85,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "ca2cefd40dd8423b8910736d376aad7a911f1e0228c64eb5ad763f0495bc793b",
        "source_sha256": {
          "third_person": "b312617b70a7a886e59947090a065c58b9b0d6cda895465ce632ce8fce451136",
          "left_wrist": "19b004bf5a8a599572e82cd414fedc6e9d31b23ccbc55cba204fce0c8d6db44e",
          "right_wrist": "e70eedbdb2c6b9aedd1b1ae57430b8ff09bb58ca9c57044879a3aa6c69a4415a"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "eb0c084d9f6a",
          "task": "robotwin/stack-bowls-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 115,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 114,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11320000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 115,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 114,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11320000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 115,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 114,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11320000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard5-20261115",
      "evaluation_date": "2026-11-15",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-rerun-kinex-tcp-posestate-06-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/06",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "06",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 37.12,
        "step_limit": 1200,
        "steps": 928,
        "success": true,
        "valid": true
      },
      "steps": 928,
      "simulation_time_s": 37.12,
      "wall_time_s": 1399.195578,
      "timing": {
        "simulation_time_s": 37.12,
        "wall_time_s": 1399.195578,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "native_instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9715521412419895,
        "cache_read_input_tokens": 1574272,
        "cache_reported_input_tokens": 1620368,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1620368,
        "cached_input_tokens": 1574272,
        "cost_usd": null,
        "fresh_input_tokens": 46096,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1620368,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1574272,
        "known_fresh_input_tokens": 46096,
        "known_input_tokens": 1620368,
        "known_output_tokens": 7915,
        "known_reasoning_output_tokens": 1442,
        "output_tokens": 7915,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1442,
        "reasoning_reported_output_tokens": 7915,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 50,
          "cache_write_input_tokens": 50,
          "cache_write_reported_input_tokens": 50,
          "cached_input_tokens": 50,
          "fresh_input_tokens": 50,
          "input_tokens": 50,
          "output_tokens": 50,
          "reasoning_output_tokens": 50,
          "reasoning_reported_output_tokens": 50
        },
        "request_attempts": 50,
        "response_count": 50,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1628283,
        "uncached_input_tokens": 46096,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11824,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11864,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11997,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12047,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12340,
            "output_tokens": 223,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 17408,
            "output_tokens": 67,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 18006,
            "output_tokens": 180,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 21922,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 23563,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 23891,
            "output_tokens": 163,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 25686,
            "output_tokens": 1253,
            "reasoning_output_tokens": 108
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26960,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27649,
            "output_tokens": 153,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 27829,
            "output_tokens": 212,
            "reasoning_output_tokens": 100
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28852,
            "output_tokens": 72,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29134,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 30163,
            "output_tokens": 809,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30993,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31387,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31783,
            "output_tokens": 215,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32177,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32568,
            "output_tokens": 205,
            "reasoning_output_tokens": 130
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33093,
            "output_tokens": 119,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33532,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33930,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34327,
            "output_tokens": 69,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 35048,
            "output_tokens": 136,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35503,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35906,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36315,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36715,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37115,
            "output_tokens": 145,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37582,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37980,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38763,
            "output_tokens": 110,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39192,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39593,
            "output_tokens": 129,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40038,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40228,
            "output_tokens": 500,
            "reasoning_output_tokens": 184
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40934,
            "output_tokens": 130,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41383,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41788,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42198,
            "output_tokens": 35,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42403,
            "output_tokens": 183,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 43293,
            "output_tokens": 110,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43720,
            "output_tokens": 142,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44176,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44404,
            "output_tokens": 89,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 45147,
            "output_tokens": 850,
            "reasoning_output_tokens": 164
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45952,
            "input_tokens": 46019,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 59,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "bash": 1,
          "exec": 33,
          "devices": 1,
          "write": 3,
          "edit": 1,
          "device": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 928 rosbridge control frames; 60 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; onboard rerun of Hard tasks 06-10.",
        "vla_contribution": {
          "progvla_exec_calls": 4,
          "direct_motion_exec_calls": 26
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-rerun-06-10-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack_blocks.md",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/workspace/.kinex/memos/stack_blocks.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/stack_blocks.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-06-20261116/workspace/.kinex/tools/stack_blocks.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 112,
        "observed_images": 29
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 60,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 372,
        "published_frames": 372,
        "decode_ok": true,
        "duration_s": 9.3,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c4333867bd8f81cd6b53ddff93f4d55d01bd469520a75766364c0b31b5d85a75",
        "source_sha256": {
          "third_person": "9639af72016f6305f4ea55f56cd7a59b895bdea08627b0e0edac0917d90c5330",
          "left_wrist": "bf231946c3f79d756cb387787dfa562153be64e693a85627934b0e8e6b2072d7",
          "right_wrist": "d7d4a9b6f9eaeb2f16cc88358ce62456e507f207a3f717b2478338d8e37ab16b"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "e76a8f984c25",
          "task": "robotwin/stack-blocks-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 373,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 372,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 37120000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 373,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 372,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 37120000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 373,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 372,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 37120000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard06-10-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-rerun-kinex-tcp-posestate-07-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/07",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "07",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 26.6,
        "step_limit": 900,
        "steps": 665,
        "success": true,
        "valid": true
      },
      "steps": 665,
      "simulation_time_s": 26.6,
      "wall_time_s": 1401.976107,
      "timing": {
        "simulation_time_s": 26.6,
        "wall_time_s": 1401.976107,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the cylindrical mug with handle on the table, rotate it, set it down in the middle, then hang it onto the rack with straight and slanted rods.",
      "native_instruction": "Grab the cylindrical mug with handle on the table, rotate it, set it down in the middle, then hang it onto the rack with straight and slanted rods.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9533418776007274,
        "cache_read_input_tokens": 2345344,
        "cache_reported_input_tokens": 2460129,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2460129,
        "cached_input_tokens": 2345344,
        "cost_usd": null,
        "fresh_input_tokens": 114785,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2460129,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2345344,
        "known_fresh_input_tokens": 114785,
        "known_input_tokens": 2460129,
        "known_output_tokens": 11627,
        "known_reasoning_output_tokens": 3741,
        "output_tokens": 11627,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3741,
        "reasoning_reported_output_tokens": 11627,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 60,
          "cache_write_input_tokens": 60,
          "cache_write_reported_input_tokens": 60,
          "cached_input_tokens": 60,
          "fresh_input_tokens": 60,
          "input_tokens": 60,
          "output_tokens": 60,
          "reasoning_output_tokens": 60,
          "reasoning_reported_output_tokens": 60
        },
        "request_attempts": 60,
        "response_count": 60,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2471756,
        "uncached_input_tokens": 114785,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11833,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11873,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12006,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15646,
            "output_tokens": 234,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 19222,
            "output_tokens": 212,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 23834,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 25475,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25803,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25994,
            "output_tokens": 150,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 28885,
            "output_tokens": 299,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29867,
            "output_tokens": 157,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30490,
            "output_tokens": 944,
            "reasoning_output_tokens": 86
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31458,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 32020,
            "output_tokens": 112,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32617,
            "output_tokens": 61,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32888,
            "output_tokens": 122,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33999,
            "output_tokens": 900,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34922,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35630,
            "output_tokens": 84,
            "reasoning_output_tokens": 6
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35743,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36142,
            "output_tokens": 98,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36601,
            "output_tokens": 175,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36998,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37396,
            "output_tokens": 292,
            "reasoning_output_tokens": 210
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38006,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38404,
            "output_tokens": 115,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38616,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39398,
            "output_tokens": 100,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39818,
            "output_tokens": 913,
            "reasoning_output_tokens": 695
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 41096,
            "output_tokens": 378,
            "reasoning_output_tokens": 294
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41793,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42584,
            "output_tokens": 118,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 43403,
            "output_tokens": 193,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 44299,
            "output_tokens": 63,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 45429,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 47291,
            "output_tokens": 122,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47442,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47360,
            "input_tokens": 48274,
            "output_tokens": 115,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 48709,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48768,
            "input_tokens": 49541,
            "output_tokens": 115,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 49974,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49920,
            "input_tokens": 50381,
            "output_tokens": 53,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 50456,
            "output_tokens": 78,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50304,
            "input_tokens": 50853,
            "output_tokens": 201,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50944,
            "input_tokens": 51282,
            "output_tokens": 114,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51200,
            "input_tokens": 51715,
            "output_tokens": 294,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50432,
            "input_tokens": 52337,
            "output_tokens": 859,
            "reasoning_output_tokens": 779
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52992,
            "input_tokens": 53264,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53248,
            "input_tokens": 53491,
            "output_tokens": 140,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53504,
            "input_tokens": 53949,
            "output_tokens": 286,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54016,
            "input_tokens": 54597,
            "output_tokens": 317,
            "reasoning_output_tokens": 222
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51840,
            "input_tokens": 55659,
            "output_tokens": 233,
            "reasoning_output_tokens": 151
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55680,
            "input_tokens": 55960,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55808,
            "input_tokens": 56375,
            "output_tokens": 160,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54784,
            "input_tokens": 57061,
            "output_tokens": 234,
            "reasoning_output_tokens": 139
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57088,
            "input_tokens": 57617,
            "output_tokens": 247,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57728,
            "input_tokens": 58006,
            "output_tokens": 65,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57856,
            "input_tokens": 58133,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56448,
            "input_tokens": 58327,
            "output_tokens": 895,
            "reasoning_output_tokens": 100
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59008,
            "input_tokens": 59247,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 73,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 19,
          "find": 1,
          "exec": 38,
          "devices": 2,
          "device": 2,
          "write": 3,
          "edit": 2,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 665 rosbridge control frames; 43 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; onboard rerun of Hard tasks 06-10.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 27
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-rerun-06-10-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/hanging_mug.md",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/workspace/.kinex/memos/hanging_mug.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-07-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 136,
        "observed_images": 23
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 43,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 267,
        "published_frames": 267,
        "decode_ok": true,
        "duration_s": 6.675,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "aca9f1c07d5179971681cbac78d80f89376f13af4823932c7790a909b0347e80",
        "source_sha256": {
          "third_person": "72912be8450921d9c5c86eb97254f358ecadea1896d794f3bacdf90b893db1f1",
          "left_wrist": "be2850fcf3d628707cb015fc9772a1dddd24e6cddb3dbb24b360c8ec1a05d1fd",
          "right_wrist": "b69b5d7656f6fcd8cb4e47c16f817d18696b1354d87c86253bcd0e9bdc53c125"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "3ee14fecbcac",
          "task": "robotwin/hanging-mug-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 268,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 267,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 26600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 268,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 267,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 26600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 268,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 267,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 26600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard06-10-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-rerun-kinex-tcp-posestate-08-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/08",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "08",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 25.2,
        "step_limit": 700,
        "steps": 630,
        "success": false,
        "valid": true
      },
      "steps": 630,
      "simulation_time_s": 25.2,
      "wall_time_s": 1412.575959,
      "timing": {
        "simulation_time_s": 25.2,
        "wall_time_s": 1412.575959,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Open the cabinet with smooth finish's drawer and set the blue soap with printed patterns inside with the other arm.",
      "native_instruction": "Open the cabinet with smooth finish's drawer and set the blue soap with printed patterns inside with the other arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9761377144523378,
        "cache_read_input_tokens": 1965056,
        "cache_reported_input_tokens": 2013093,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2013093,
        "cached_input_tokens": 1965056,
        "cost_usd": null,
        "fresh_input_tokens": 48037,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2013093,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1965056,
        "known_fresh_input_tokens": 48037,
        "known_input_tokens": 2013093,
        "known_output_tokens": 11067,
        "known_reasoning_output_tokens": 4672,
        "output_tokens": 11067,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 4672,
        "reasoning_reported_output_tokens": 11067,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 59,
          "cache_write_input_tokens": 59,
          "cache_write_reported_input_tokens": 59,
          "cached_input_tokens": 59,
          "fresh_input_tokens": 59,
          "input_tokens": 59,
          "output_tokens": 59,
          "reasoning_output_tokens": 59,
          "reasoning_reported_output_tokens": 59
        },
        "request_attempts": 59,
        "response_count": 59,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2024160,
        "uncached_input_tokens": 48037,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11825,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11958,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12006,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12299,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12762,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12672,
            "input_tokens": 13576,
            "output_tokens": 195,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13568,
            "input_tokens": 18201,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18771,
            "output_tokens": 150,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 21865,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 23265,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23593,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23782,
            "output_tokens": 299,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24497,
            "output_tokens": 164,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25127,
            "output_tokens": 962,
            "reasoning_output_tokens": 158
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26111,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 27336,
            "output_tokens": 71,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27469,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27721,
            "output_tokens": 71,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28971,
            "output_tokens": 1046,
            "reasoning_output_tokens": 294
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30038,
            "output_tokens": 117,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30534,
            "output_tokens": 375,
            "reasoning_output_tokens": 260
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31280,
            "output_tokens": 92,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31434,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31795,
            "output_tokens": 237,
            "reasoning_output_tokens": 159
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32895,
            "output_tokens": 167,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33437,
            "output_tokens": 205,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34339,
            "output_tokens": 177,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 35213,
            "output_tokens": 150,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35736,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36098,
            "output_tokens": 349,
            "reasoning_output_tokens": 239
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36821,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37012,
            "output_tokens": 137,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37521,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37976,
            "output_tokens": 455,
            "reasoning_output_tokens": 396
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38808,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39365,
            "output_tokens": 129,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39866,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40314,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40505,
            "output_tokens": 343,
            "reasoning_output_tokens": 204
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41045,
            "output_tokens": 209,
            "reasoning_output_tokens": 102
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41627,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41987,
            "output_tokens": 208,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42568,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42929,
            "output_tokens": 378,
            "reasoning_output_tokens": 303
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43681,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 43873,
            "output_tokens": 270,
            "reasoning_output_tokens": 193
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44518,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44710,
            "output_tokens": 156,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 45239,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45432,
            "output_tokens": 382,
            "reasoning_output_tokens": 296
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 46186,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46379,
            "output_tokens": 314,
            "reasoning_output_tokens": 197
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 47064,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47425,
            "output_tokens": 347,
            "reasoning_output_tokens": 228
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 48144,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48000,
            "input_tokens": 48505,
            "output_tokens": 892,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49280,
            "input_tokens": 49425,
            "output_tokens": 354,
            "reasoning_output_tokens": 309
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49664,
            "input_tokens": 49841,
            "output_tokens": 530,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50176,
            "input_tokens": 50393,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 66,
        "model_tool_calls_by_name": {
          "read": 17,
          "ls": 1,
          "bash": 1,
          "exec": 26,
          "devices": 7,
          "device": 7,
          "write": 3,
          "device_stop": 3,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 630 rosbridge control frames; 30 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; onboard rerun of Hard tasks 06-10.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 21
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-rerun-06-10-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/cabinet.md",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/workspace/.kinex/memos/cabinet.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-08-20261116/workspace/.kinex/tools/tcp_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 128,
        "observed_images": 41
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 30,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 253,
        "published_frames": 253,
        "decode_ok": true,
        "duration_s": 6.325,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "a6ff42f983c230fbdf455ba4335f638fe8b7f8ca8dbd51149e79d7addb8ba0dd",
        "source_sha256": {
          "third_person": "d64e7d067c8fe8083af80ea7ead4abb0ef2548bd35e1e4cca77520b6ed6928b0",
          "left_wrist": "793f2de3f593a04feb6b30c2af4011795e22b30a65691e1f64533c66351bff42",
          "right_wrist": "c560784cafc208cf630b79ef7bdf8de619497b0752e51d286a711c554dea37e6"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "216471a3a7b4",
          "task": "robotwin/put-object-cabinet-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 254,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 253,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 25200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 254,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 253,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 25200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 254,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 253,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 25200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard06-10-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-rerun-kinex-tcp-posestate-09-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/09",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "09",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 37.04,
        "step_limit": 1200,
        "steps": 926,
        "success": true,
        "valid": true
      },
      "steps": 926,
      "simulation_time_s": 37.04,
      "wall_time_s": 1249.634128,
      "timing": {
        "simulation_time_s": 37.04,
        "wall_time_s": 1249.634128,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move large block, medium block, and small block to the center table.",
      "native_instruction": "Move large block, medium block, and small block to the center table.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9242867452781557,
        "cache_read_input_tokens": 914432,
        "cache_reported_input_tokens": 989338,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 989338,
        "cached_input_tokens": 914432,
        "cost_usd": null,
        "fresh_input_tokens": 74906,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 989338,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 914432,
        "known_fresh_input_tokens": 74906,
        "known_input_tokens": 989338,
        "known_output_tokens": 7033,
        "known_reasoning_output_tokens": 1332,
        "output_tokens": 7033,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1332,
        "reasoning_reported_output_tokens": 7033,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 35,
          "cache_write_input_tokens": 35,
          "cache_write_reported_input_tokens": 35,
          "cached_input_tokens": 35,
          "fresh_input_tokens": 35,
          "input_tokens": 35,
          "output_tokens": 35,
          "reasoning_output_tokens": 35,
          "reasoning_reported_output_tokens": 35
        },
        "request_attempts": 35,
        "response_count": 35,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 996371,
        "uncached_input_tokens": 74906,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11821,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11861,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11994,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12042,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12335,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18225,
            "output_tokens": 137,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 20431,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20759,
            "output_tokens": 162,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21317,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 21725,
            "output_tokens": 158,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 23614,
            "output_tokens": 171,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24824,
            "output_tokens": 383,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25557,
            "output_tokens": 330,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26114,
            "output_tokens": 1149,
            "reasoning_output_tokens": 289
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27285,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 27882,
            "output_tokens": 188,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 30424,
            "output_tokens": 1028,
            "reasoning_output_tokens": 158
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31473,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31943,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 32415,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32904,
            "output_tokens": 186,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33480,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34004,
            "output_tokens": 214,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34607,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35094,
            "output_tokens": 124,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 35608,
            "output_tokens": 134,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 36125,
            "output_tokens": 145,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36660,
            "output_tokens": 211,
            "reasoning_output_tokens": 170
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36893,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37383,
            "output_tokens": 161,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37930,
            "output_tokens": 134,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38206,
            "output_tokens": 53,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38429,
            "output_tokens": 89,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38652,
            "output_tokens": 647,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39322,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 46,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 18,
          "bash": 1,
          "devices": 2,
          "exec": 18,
          "write": 3,
          "device_stop": 1,
          "device": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 926 rosbridge control frames; 50 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; onboard rerun of Hard tasks 06-10.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 13
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-rerun-06-10-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/blocks-ranking-size.md",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/workspace/.kinex/memos/blocks-ranking-size.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-09-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 84,
        "observed_images": 15
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 50,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 372,
        "published_frames": 372,
        "decode_ok": true,
        "duration_s": 9.3,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c1c5eec6f4955f13bd042ded7b324c4c863d51eb2a2679987e30d1c776911d70",
        "source_sha256": {
          "third_person": "f17c2784fc1b382ff97903c92341ae9c33ab1abfdc762104b8b23457ead66bda",
          "left_wrist": "ac06aa57478dc1bb48efd8c09f3495183ad618641ada9b21f9fcea84d68eecc1",
          "right_wrist": "168ade248b59ffe0f02cee4a07392eb2309d945276c1c7a96af1e13213d677da"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "dca755a05cc2",
          "task": "robotwin/blocks-ranking-size-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 372,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 372,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 37040000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 372,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 372,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 37040000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 372,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 372,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 37040000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard06-10-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-rerun-kinex-tcp-posestate-10-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/10",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "10",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.92,
        "step_limit": 400,
        "steps": 248,
        "success": true,
        "valid": true
      },
      "steps": 248,
      "simulation_time_s": 9.92,
      "wall_time_s": 885.72805,
      "timing": {
        "simulation_time_s": 9.92,
        "wall_time_s": 885.72805,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Hold the dark gray cooking kitchenpot firmly and lift it",
      "native_instruction": "Hold the dark gray cooking kitchenpot firmly and lift it",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9538640030977088,
        "cache_read_input_tokens": 662656,
        "cache_reported_input_tokens": 694707,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 694707,
        "cached_input_tokens": 662656,
        "cost_usd": null,
        "fresh_input_tokens": 32051,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 694707,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 662656,
        "known_fresh_input_tokens": 32051,
        "known_input_tokens": 694707,
        "known_output_tokens": 5463,
        "known_reasoning_output_tokens": 770,
        "output_tokens": 5463,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 770,
        "reasoning_reported_output_tokens": 5463,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 28,
          "cache_write_input_tokens": 28,
          "cache_write_reported_input_tokens": 28,
          "cached_input_tokens": 28,
          "fresh_input_tokens": 28,
          "input_tokens": 28,
          "output_tokens": 28,
          "reasoning_output_tokens": 28,
          "reasoning_reported_output_tokens": 28
        },
        "request_attempts": 28,
        "response_count": 28,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 700170,
        "uncached_input_tokens": 32051,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11810,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11850,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11983,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12031,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13217,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13312,
            "input_tokens": 17474,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 21290,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21618,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 23018,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23540,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23948,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24498,
            "output_tokens": 100,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25064,
            "output_tokens": 264,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25787,
            "output_tokens": 1099,
            "reasoning_output_tokens": 235
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26908,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27043,
            "output_tokens": 117,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 28103,
            "output_tokens": 125,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28428,
            "output_tokens": 358,
            "reasoning_output_tokens": 212
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28892,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29968,
            "output_tokens": 97,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30275,
            "output_tokens": 884,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31181,
            "output_tokens": 115,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31857,
            "output_tokens": 244,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32336,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32606,
            "output_tokens": 43,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32711,
            "output_tokens": 187,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33349,
            "output_tokens": 551,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 33922,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 36,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 15,
          "find": 1,
          "devices": 1,
          "exec": 13,
          "write": 3,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 248 rosbridge control frames; 45 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; onboard rerun of Hard tasks 06-10.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 3
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-rerun-06-10-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/lift-pot.md",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/workspace/.kinex/memos/lift-pot.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_dual.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/workspace/.kinex/tools/cartesian_dual.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-rerun-06-10-20261116/onboard-rerun-kinex-tcp-posestate-10-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 67,
        "observed_images": 12
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 45,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 100,
        "published_frames": 100,
        "decode_ok": true,
        "duration_s": 2.5,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "9db0854589c0fd26df9b836ad45b1b58d104bacbcb9a607fe3bb0f6ad27edde7",
        "source_sha256": {
          "third_person": "252ded1fc0f5380a46e21912fc655e60a71ad5d108e49d23acd7ae3b7e9e304a",
          "left_wrist": "9297b5a79070e9b21b7e074f83f55f98fcc33f45feb465b2a80546126b4738f3",
          "right_wrist": "a822b9d1512c81fded0caf5b9b4dbdbd33699c7a44e2d18dd4e896c0e4405ca8"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "7bc61d720798",
          "task": "robotwin/lift-pot-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 101,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 100,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9920000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 101,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 100,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9920000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 101,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 100,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9920000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard06-10-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-11-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/11",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "11",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 2
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 14.200000000000001,
        "step_limit": 400,
        "steps": 355,
        "success": false,
        "valid": true
      },
      "steps": 355,
      "simulation_time_s": 14.200000000000001,
      "wall_time_s": 871.072208,
      "timing": {
        "simulation_time_s": 14.200000000000001,
        "wall_time_s": 871.072208,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the green plastic bottle with ridged bottom carefully using the right arm",
      "native_instruction": "Pick up the green plastic bottle with ridged bottom carefully using the right arm",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9586976546842114,
        "cache_read_input_tokens": 858624,
        "cache_reported_input_tokens": 895615,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 895615,
        "cached_input_tokens": 858624,
        "cost_usd": null,
        "fresh_input_tokens": 36991,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 895615,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 858624,
        "known_fresh_input_tokens": 36991,
        "known_input_tokens": 895615,
        "known_output_tokens": 6535,
        "known_reasoning_output_tokens": 1485,
        "output_tokens": 6535,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1485,
        "reasoning_reported_output_tokens": 6535,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 34,
          "cache_write_input_tokens": 34,
          "cache_write_reported_input_tokens": 34,
          "cached_input_tokens": 34,
          "fresh_input_tokens": 34,
          "input_tokens": 34,
          "output_tokens": 34,
          "reasoning_output_tokens": 34,
          "reasoning_reported_output_tokens": 34
        },
        "request_attempts": 34,
        "response_count": 34,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 902150,
        "uncached_input_tokens": 36991,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11988,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12038,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12331,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 13148,
            "output_tokens": 163,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13184,
            "input_tokens": 16599,
            "output_tokens": 195,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 20430,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20758,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 20948,
            "output_tokens": 112,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 22397,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22805,
            "output_tokens": 257,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23699,
            "output_tokens": 74,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24239,
            "output_tokens": 940,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25201,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25364,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 28189,
            "output_tokens": 172,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28571,
            "output_tokens": 370,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29484,
            "output_tokens": 37,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29730,
            "output_tokens": 53,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30772,
            "output_tokens": 753,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31546,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32106,
            "output_tokens": 164,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32744,
            "output_tokens": 79,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32845,
            "output_tokens": 262,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33450,
            "output_tokens": 196,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34122,
            "output_tokens": 194,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34538,
            "output_tokens": 133,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35145,
            "output_tokens": 623,
            "reasoning_output_tokens": 509
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36269,
            "output_tokens": 93,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36831,
            "output_tokens": 282,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37518,
            "output_tokens": 256,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37802,
            "output_tokens": 513,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38338,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 41,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "bash": 1,
          "devices": 1,
          "device": 2,
          "exec": 14,
          "write": 3,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 355 rosbridge control frames; 25 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 7
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/bottle_pickup.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/workspace/.kinex/memos/bottle_pickup.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_right.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/workspace/.kinex/tools/cartesian_right.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-11-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 78,
        "observed_images": 15
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 25,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 143,
        "published_frames": 143,
        "decode_ok": true,
        "duration_s": 3.575,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "92a65e2e997ca1c10c0fbc62f5bf334792e0b1c71c83505befce788ea75e7701",
        "source_sha256": {
          "third_person": "cccf2eea8271e933429bab30010d9dcbc72f2da9d32ee7fbef64c5074004ee92",
          "left_wrist": "f727debdbc4d79c7bd2a5a862290dccbd1c96131432e17ec42c670e3ed0cbc46",
          "right_wrist": "83fbccfdcb841ae755d6a7e97beb1cb137bd437fd5d64b28d4f7d6e8d1e82568"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "0ece481ce550",
          "task": "robotwin/adjust-bottle-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 144,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 143,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 144,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 143,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 144,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 143,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-12-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/12",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "12",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.92,
        "step_limit": 400,
        "steps": 248,
        "success": true,
        "valid": true
      },
      "steps": 248,
      "simulation_time_s": 9.92,
      "wall_time_s": 764.02249,
      "timing": {
        "simulation_time_s": 9.92,
        "wall_time_s": 764.02249,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the hammer with black handle and hit the block",
      "native_instruction": "Grab the hammer with black handle and hit the block",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9294491411340262,
        "cache_read_input_tokens": 1035648,
        "cache_reported_input_tokens": 1114260,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1114260,
        "cached_input_tokens": 1035648,
        "cost_usd": null,
        "fresh_input_tokens": 78612,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1114260,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1035648,
        "known_fresh_input_tokens": 78612,
        "known_input_tokens": 1114260,
        "known_output_tokens": 6441,
        "known_reasoning_output_tokens": 1723,
        "output_tokens": 6441,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1723,
        "reasoning_reported_output_tokens": 6441,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 38,
          "cache_write_input_tokens": 38,
          "cache_write_reported_input_tokens": 38,
          "cached_input_tokens": 38,
          "fresh_input_tokens": 38,
          "input_tokens": 38,
          "output_tokens": 38,
          "reasoning_output_tokens": 38,
          "reasoning_reported_output_tokens": 38
        },
        "request_attempts": 38,
        "response_count": 38,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1120701,
        "uncached_input_tokens": 78612,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11812,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11852,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11985,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12035,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12709,
            "output_tokens": 214,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12800,
            "input_tokens": 16253,
            "output_tokens": 212,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 20865,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 21193,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 22593,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22784,
            "output_tokens": 86,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23909,
            "output_tokens": 270,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24574,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24982,
            "output_tokens": 270,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25661,
            "output_tokens": 119,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 26246,
            "output_tokens": 887,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27157,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 31540,
            "output_tokens": 94,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 31844,
            "output_tokens": 96,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 33119,
            "output_tokens": 751,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 33893,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 34275,
            "output_tokens": 91,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34664,
            "output_tokens": 49,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34764,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34903,
            "output_tokens": 75,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34996,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35078,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35460,
            "output_tokens": 156,
            "reasoning_output_tokens": 131
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35785,
            "output_tokens": 163,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 36643,
            "output_tokens": 595,
            "reasoning_output_tokens": 514
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 37303,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37706,
            "output_tokens": 127,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 38156,
            "output_tokens": 284,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38704,
            "output_tokens": 398,
            "reasoning_output_tokens": 274
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 39244,
            "output_tokens": 58,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39364,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39794,
            "output_tokens": 101,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39925,
            "output_tokens": 542,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40490,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 48,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 18,
          "find": 1,
          "devices": 3,
          "exec": 11,
          "device": 5,
          "write": 3,
          "relative_pose": 1,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 248 rosbridge control frames; 20 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 5
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/hammer.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/workspace/.kinex/memos/hammer.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-12-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 89,
        "observed_images": 13
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 20,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 100,
        "published_frames": 100,
        "decode_ok": true,
        "duration_s": 2.5,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "4320f18db8d9e40b4a0991d21d6799bd11504de5133c8a3ce9fa417c239534fe",
        "source_sha256": {
          "third_person": "2da0dfe4bc84c4b48fe90b04a53bfb721cacd54b60b2ec4681ae21a58403ce11",
          "left_wrist": "e4ed902318682d23fe418a475a5bf874d99d6e2f0cf2e99168d1c55896273a9a",
          "right_wrist": "9c7af9df879c0ab21131546fb7024989dd59e88d28d4c0ea2bd88a2f9a99484d"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "c96d33a789b1",
          "task": "robotwin/beat-block-hammer-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 101,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 100,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9920000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 101,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 100,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9920000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 101,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 100,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9920000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-13-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/13",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "13",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 31.28,
        "step_limit": 1200,
        "steps": 782,
        "success": true,
        "valid": true
      },
      "steps": 782,
      "simulation_time_s": 31.28,
      "wall_time_s": 793.007644,
      "timing": {
        "simulation_time_s": 31.28,
        "wall_time_s": 793.007644,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set red block on the left, then green block in the center, and finally blue block on the right.",
      "native_instruction": "Set red block on the left, then green block in the center, and finally blue block on the right.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9089689018476628,
        "cache_read_input_tokens": 730112,
        "cache_reported_input_tokens": 803231,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 803231,
        "cached_input_tokens": 730112,
        "cost_usd": null,
        "fresh_input_tokens": 73119,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 803231,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 730112,
        "known_fresh_input_tokens": 73119,
        "known_input_tokens": 803231,
        "known_output_tokens": 6414,
        "known_reasoning_output_tokens": 1452,
        "output_tokens": 6414,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1452,
        "reasoning_reported_output_tokens": 6414,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 29,
          "cache_write_input_tokens": 29,
          "cache_write_reported_input_tokens": 29,
          "cached_input_tokens": 29,
          "fresh_input_tokens": 29,
          "input_tokens": 29,
          "output_tokens": 29,
          "reasoning_output_tokens": 29,
          "reasoning_reported_output_tokens": 29
        },
        "request_attempts": 29,
        "response_count": 29,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 809645,
        "uncached_input_tokens": 73119,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11826,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11866,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11999,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 14400,
            "output_tokens": 229,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 18716,
            "output_tokens": 155,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 20988,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 21218,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21407,
            "output_tokens": 277,
            "reasoning_output_tokens": 146
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 23298,
            "output_tokens": 386,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24685,
            "output_tokens": 158,
            "reasoning_output_tokens": 118
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 25309,
            "output_tokens": 515,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26401,
            "output_tokens": 247,
            "reasoning_output_tokens": 175
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27827,
            "output_tokens": 942,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 28792,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29282,
            "output_tokens": 125,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 30019,
            "output_tokens": 486,
            "reasoning_output_tokens": 208
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30655,
            "output_tokens": 250,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31727,
            "output_tokens": 199,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32744,
            "output_tokens": 134,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33879,
            "output_tokens": 105,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34440,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 34934,
            "output_tokens": 101,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 35651,
            "output_tokens": 172,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35965,
            "output_tokens": 81,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36108,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 36301,
            "output_tokens": 657,
            "reasoning_output_tokens": 246
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 37216,
            "output_tokens": 261,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37506,
            "output_tokens": 542,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38072,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 40,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 15,
          "find": 1,
          "devices": 1,
          "device": 2,
          "exec": 13,
          "write": 2,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 782 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 8
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/blocks-ranking.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/workspace/.kinex/memos/blocks-ranking.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-13-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 72,
        "observed_images": 9
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 314,
        "published_frames": 314,
        "decode_ok": true,
        "duration_s": 7.85,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "e23659ed4aa74238f12925aa451841227a79ed94fa00e253c6cb56eddf39b587",
        "source_sha256": {
          "third_person": "3d6a842b92b79a61d6ca8eb38e9a704f9e46a87410b41baae651feebdc5d74a7",
          "left_wrist": "c7af417bd3cf1255148dd69dc74f038d83774ef665784f26e3665349c2d0e17e",
          "right_wrist": "0f9563c44a7ade607c8c402a335503a94060cbbb1d0f923cc42200bb6e0bfac2"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "9366f2d60be8",
          "task": "robotwin/blocks-ranking-rgb-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 314,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 314,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 31280000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 314,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 314,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 31280000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 314,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 314,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 31280000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": false
    },
    {
      "id": "onboard-kinex-tcp-posestate-14-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/14",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "14",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 5.5600000000000005,
        "step_limit": 400,
        "steps": 139,
        "success": true,
        "valid": true
      },
      "steps": 139,
      "simulation_time_s": 5.5600000000000005,
      "wall_time_s": 599.510202,
      "timing": {
        "simulation_time_s": 5.5600000000000005,
        "wall_time_s": 599.510202,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Click the top button of the alarm-clock with beige screen",
      "native_instruction": "Click the top button of the alarm-clock with beige screen",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.931165436401598,
        "cache_read_input_tokens": 679680,
        "cache_reported_input_tokens": 729924,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 729924,
        "cached_input_tokens": 679680,
        "cost_usd": null,
        "fresh_input_tokens": 50244,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 729924,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 679680,
        "known_fresh_input_tokens": 50244,
        "known_input_tokens": 729924,
        "known_output_tokens": 4447,
        "known_reasoning_output_tokens": 495,
        "output_tokens": 4447,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 495,
        "reasoning_reported_output_tokens": 4447,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 25,
          "cache_write_input_tokens": 25,
          "cache_write_reported_input_tokens": 25,
          "cached_input_tokens": 25,
          "fresh_input_tokens": 25,
          "input_tokens": 25,
          "output_tokens": 25,
          "reasoning_output_tokens": 25,
          "reasoning_reported_output_tokens": 25
        },
        "request_attempts": 25,
        "response_count": 25,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 734371,
        "uncached_input_tokens": 50244,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11814,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11854,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11987,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15722,
            "output_tokens": 247,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 20068,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20096,
            "input_tokens": 23884,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24212,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 26055,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26246,
            "output_tokens": 80,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26695,
            "output_tokens": 303,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 28411,
            "output_tokens": 74,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28807,
            "output_tokens": 887,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29717,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 34032,
            "output_tokens": 60,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34302,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 35331,
            "output_tokens": 97,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35894,
            "output_tokens": 267,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36416,
            "output_tokens": 705,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37142,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37791,
            "output_tokens": 220,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38149,
            "output_tokens": 67,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38278,
            "output_tokens": 131,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38764,
            "output_tokens": 151,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 38942,
            "output_tokens": 446,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39411,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 36,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 16,
          "find": 1,
          "devices": 2,
          "exec": 7,
          "device": 1,
          "write": 3,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 139 rosbridge control frames; 20 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 3
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/click_alarmclock.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/workspace/.kinex/memos/click_alarmclock.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-14-20261116/workspace/.kinex/tools/tcp_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 64,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 20,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 57,
        "published_frames": 57,
        "decode_ok": true,
        "duration_s": 1.425,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "4b1ceebbc3e95f490e26334b448fab48b9e1757a04dfe60563930917c9b8bb1e",
        "source_sha256": {
          "third_person": "65a2b8374c0e2c773b2dd5b15fb6f67c97781270a141655e20266c31beab3d70",
          "left_wrist": "978f77cec4ff06062a288196710627a437af405bdefac3873a40401f2e29ef46",
          "right_wrist": "3b2d86ab4a991df2bf2570e7810d869fda5ba2fd0529cf59c0c888f3d0a7ec1e"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "784ef56ca8d7",
          "task": "robotwin/click-alarmclock-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 57,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 57,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 5560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 57,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 57,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 5560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 57,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 57,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 5560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-15-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/15",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "15",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 14.200000000000001,
        "step_limit": 400,
        "steps": 355,
        "success": false,
        "valid": true
      },
      "steps": 355,
      "simulation_time_s": 14.200000000000001,
      "wall_time_s": 662.74652,
      "timing": {
        "simulation_time_s": 14.200000000000001,
        "wall_time_s": 662.74652,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to press the medium-sized bell's top center.",
      "native_instruction": "Use the left arm to press the medium-sized bell's top center.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9631069198927494,
        "cache_read_input_tokens": 1208704,
        "cache_reported_input_tokens": 1255005,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1255005,
        "cached_input_tokens": 1208704,
        "cost_usd": null,
        "fresh_input_tokens": 46301,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1255005,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1208704,
        "known_fresh_input_tokens": 46301,
        "known_input_tokens": 1255005,
        "known_output_tokens": 6868,
        "known_reasoning_output_tokens": 2147,
        "output_tokens": 6868,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2147,
        "reasoning_reported_output_tokens": 6868,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 43,
          "cache_write_input_tokens": 43,
          "cache_write_reported_input_tokens": 43,
          "cached_input_tokens": 43,
          "fresh_input_tokens": 43,
          "input_tokens": 43,
          "output_tokens": 43,
          "reasoning_output_tokens": 43,
          "reasoning_reported_output_tokens": 43
        },
        "request_attempts": 43,
        "response_count": 43,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1261873,
        "uncached_input_tokens": 46301,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11816,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11856,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11989,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12037,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12330,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16587,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 17849,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18419,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 18583,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 18772,
            "output_tokens": 212,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 21769,
            "output_tokens": 95,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 21972,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22523,
            "output_tokens": 239,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 23619,
            "output_tokens": 228,
            "reasoning_output_tokens": 186
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24836,
            "output_tokens": 858,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25715,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26509,
            "output_tokens": 106,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 27447,
            "output_tokens": 110,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28027,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28083,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28164,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28706,
            "output_tokens": 357,
            "reasoning_output_tokens": 120
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29421,
            "output_tokens": 133,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30249,
            "output_tokens": 178,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31151,
            "output_tokens": 156,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31375,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 32278,
            "output_tokens": 99,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32778,
            "output_tokens": 286,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33776,
            "output_tokens": 76,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34318,
            "output_tokens": 144,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 35669,
            "output_tokens": 158,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 37039,
            "output_tokens": 47,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37255,
            "output_tokens": 94,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37422,
            "output_tokens": 136,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38821,
            "output_tokens": 612,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 40697,
            "output_tokens": 176,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 42010,
            "output_tokens": 300,
            "reasoning_output_tokens": 205
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 43571,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 44470,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 45778,
            "output_tokens": 214,
            "reasoning_output_tokens": 179
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46014,
            "output_tokens": 364,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46405,
            "output_tokens": 472,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 46900,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 50,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 15,
          "bash": 1,
          "exec": 19,
          "devices": 4,
          "device": 3,
          "write": 2,
          "device_stop": 2,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 355 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 13
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/click-bell.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/workspace/.kinex/memos/click-bell.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/left_tcp.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-15-20261116/workspace/.kinex/tools/left_tcp.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 96,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 143,
        "published_frames": 143,
        "decode_ok": true,
        "duration_s": 3.575,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "d03de51330f83e4c9ac153d7fb1062e524074b13c864424ef92bebb7e2d4f14d",
        "source_sha256": {
          "third_person": "678f1bf5429b3c7de72ff91c6c1f6991af3b737e2d889d4b3f577f1fc5003d58",
          "left_wrist": "a55401d7a352a0520e17b2bacfcf483d8f75a8128283abefef7becb9aeac79af",
          "right_wrist": "90820f63ed8e33b941012ba493ba3fcb425b6093d303b429112f97e304c21544"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "f61621f645c5",
          "task": "robotwin/click-bell-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 144,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 143,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 144,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 143,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 144,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 143,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": false
    },
    {
      "id": "onboard-kinex-tcp-posestate-16-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/16",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "16",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 19.080000000000002,
        "step_limit": 600,
        "steps": 477,
        "success": false,
        "valid": true
      },
      "steps": 477,
      "simulation_time_s": 19.080000000000002,
      "wall_time_s": 1265.743171,
      "timing": {
        "simulation_time_s": 19.080000000000002,
        "wall_time_s": 1265.743171,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the compact tabletop trashbin, pour balls into the bin.",
      "native_instruction": "Pick up the compact tabletop trashbin, pour balls into the bin.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9531292977390454,
        "cache_read_input_tokens": 1490048,
        "cache_reported_input_tokens": 1563322,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1563322,
        "cached_input_tokens": 1490048,
        "cost_usd": null,
        "fresh_input_tokens": 73274,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1563322,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1490048,
        "known_fresh_input_tokens": 73274,
        "known_input_tokens": 1563322,
        "known_output_tokens": 11221,
        "known_reasoning_output_tokens": 4724,
        "output_tokens": 11221,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 4724,
        "reasoning_reported_output_tokens": 11221,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 46,
          "cache_write_input_tokens": 46,
          "cache_write_reported_input_tokens": 46,
          "cached_input_tokens": 46,
          "fresh_input_tokens": 46,
          "input_tokens": 46,
          "output_tokens": 46,
          "reasoning_output_tokens": 46,
          "reasoning_reported_output_tokens": 46
        },
        "request_attempts": 46,
        "response_count": 46,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1574543,
        "uncached_input_tokens": 73274,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11817,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11857,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11990,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 12040,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13244,
            "output_tokens": 181,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13312,
            "input_tokens": 15966,
            "output_tokens": 182,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 19799,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 23033,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23361,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23551,
            "output_tokens": 61,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23933,
            "output_tokens": 174,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 27101,
            "output_tokens": 310,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 29254,
            "output_tokens": 1159,
            "reasoning_output_tokens": 318
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30437,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 31255,
            "output_tokens": 83,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31400,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31652,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32871,
            "output_tokens": 1067,
            "reasoning_output_tokens": 376
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 33961,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34547,
            "output_tokens": 98,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 35145,
            "output_tokens": 81,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35248,
            "output_tokens": 107,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35821,
            "output_tokens": 341,
            "reasoning_output_tokens": 185
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36382,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36959,
            "output_tokens": 418,
            "reasoning_output_tokens": 160
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37739,
            "output_tokens": 434,
            "reasoning_output_tokens": 371
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38295,
            "output_tokens": 147,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38943,
            "output_tokens": 154,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39318,
            "output_tokens": 169,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39986,
            "output_tokens": 50,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40058,
            "output_tokens": 610,
            "reasoning_output_tokens": 413
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40930,
            "output_tokens": 259,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41295,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41874,
            "output_tokens": 572,
            "reasoning_output_tokens": 291
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42784,
            "output_tokens": 283,
            "reasoning_output_tokens": 199
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 43565,
            "output_tokens": 46,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 43673,
            "output_tokens": 552,
            "reasoning_output_tokens": 475
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44909,
            "output_tokens": 106,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45506,
            "output_tokens": 129,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 45657,
            "output_tokens": 142,
            "reasoning_output_tokens": 116
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 46007,
            "output_tokens": 748,
            "reasoning_output_tokens": 489
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 47091,
            "output_tokens": 65,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 47218,
            "output_tokens": 1013,
            "reasoning_output_tokens": 661
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48128,
            "input_tokens": 48260,
            "output_tokens": 232,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 48522,
            "output_tokens": 522,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49068,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 57,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 17,
          "find": 1,
          "exec": 19,
          "devices": 2,
          "device": 2,
          "write": 3,
          "device_stop": 6,
          "grep": 1,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 477 rosbridge control frames; 30 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 13
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/dump_bin.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/workspace/.kinex/memos/dump_bin.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-16-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 106,
        "observed_images": 25
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 30,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 192,
        "published_frames": 192,
        "decode_ok": true,
        "duration_s": 4.8,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "9c693bde6db46fc2e087c87216c4bfa84f3cd584f0c136a2dbdf76cb32ffa5ca",
        "source_sha256": {
          "third_person": "cc8725b37d7bf1e89b4b25e7e4f988e471235efa1c62e72d1466542b961dd453",
          "left_wrist": "402f2e8b4a55cb2e8893ee54ae260cb33bab18813251a95ec7d58b52e55142cd",
          "right_wrist": "ef9692c1a10e2ceef86238e0dc44347aa47e99feade8eb92cc7c9f911fe7d936"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "4d8601ee60d3",
          "task": "robotwin/dump-bin-bigbin-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 192,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 192,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19080000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 192,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 192,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19080000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 192,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 192,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19080000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-17-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/17",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "17",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.040000000000001,
        "step_limit": 400,
        "steps": 301,
        "success": true,
        "valid": true
      },
      "steps": 301,
      "simulation_time_s": 12.040000000000001,
      "wall_time_s": 826.98506,
      "timing": {
        "simulation_time_s": 12.040000000000001,
        "wall_time_s": 826.98506,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use both arms to grab the light brown wooden roller.",
      "native_instruction": "Use both arms to grab the light brown wooden roller.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9502964546797802,
        "cache_read_input_tokens": 759552,
        "cache_reported_input_tokens": 799279,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 799279,
        "cached_input_tokens": 759552,
        "cost_usd": null,
        "fresh_input_tokens": 39727,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 799279,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 759552,
        "known_fresh_input_tokens": 39727,
        "known_input_tokens": 799279,
        "known_output_tokens": 5986,
        "known_reasoning_output_tokens": 896,
        "output_tokens": 5986,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 896,
        "reasoning_reported_output_tokens": 5986,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 28,
          "cache_write_input_tokens": 28,
          "cache_write_reported_input_tokens": 28,
          "cached_input_tokens": 28,
          "fresh_input_tokens": 28,
          "input_tokens": 28,
          "output_tokens": 28,
          "reasoning_output_tokens": 28,
          "reasoning_reported_output_tokens": 28
        },
        "request_attempts": 28,
        "response_count": 28,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 805265,
        "uncached_input_tokens": 39727,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11811,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11851,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11984,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 15568,
            "output_tokens": 247,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15616,
            "input_tokens": 19939,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 23730,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24058,
            "output_tokens": 161,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24614,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25022,
            "output_tokens": 161,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26915,
            "output_tokens": 335,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27571,
            "output_tokens": 100,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28137,
            "output_tokens": 1010,
            "reasoning_output_tokens": 152
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29169,
            "output_tokens": 101,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29335,
            "output_tokens": 185,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30412,
            "output_tokens": 217,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 31525,
            "output_tokens": 104,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31839,
            "output_tokens": 389,
            "reasoning_output_tokens": 152
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32455,
            "output_tokens": 118,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33752,
            "output_tokens": 863,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34637,
            "output_tokens": 137,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35147,
            "output_tokens": 143,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35660,
            "output_tokens": 123,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36157,
            "output_tokens": 197,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36683,
            "output_tokens": 156,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 36978,
            "output_tokens": 188,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37815,
            "output_tokens": 184,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38027,
            "output_tokens": 438,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38488,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 39,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 16,
          "find": 1,
          "devices": 1,
          "exec": 13,
          "write": 3,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 301 rosbridge control frames; 44 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 5
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/grab_roller.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/workspace/.kinex/memos/grab_roller.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_dual.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/workspace/.kinex/tools/cartesian_dual.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-17-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 70,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 44,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 122,
        "published_frames": 122,
        "decode_ok": true,
        "duration_s": 3.05,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "2bf3f908981abc020cad854d206bd8cd235373aecde210416af330603e0b89ea",
        "source_sha256": {
          "third_person": "407f62839505b1494400cb69d5ac5f56fc32daf03ecd47966be4d22fc2bcff75",
          "left_wrist": "2332688747998ca5ec297b44e58724319982e8cf81989af5c1dd86881280fb6e",
          "right_wrist": "a4bc6336258a125c4ab8147db56ed5cc01c5634bdddef997b77b43fe69e4d60d"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "3ae56771615d",
          "task": "robotwin/grab-roller-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 122,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 122,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12040000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 122,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 122,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12040000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 122,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 122,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12040000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-18-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/18",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "18",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 23.12,
        "step_limit": 600,
        "steps": 578,
        "success": true,
        "valid": true
      },
      "steps": 578,
      "simulation_time_s": 23.12,
      "wall_time_s": 1353.911674,
      "timing": {
        "simulation_time_s": 23.12,
        "wall_time_s": 1353.911674,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the microphone used for audio input and pass it to the other arm.",
      "native_instruction": "Grab the microphone used for audio input and pass it to the other arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9745306203135038,
        "cache_read_input_tokens": 1735296,
        "cache_reported_input_tokens": 1780648,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1780648,
        "cached_input_tokens": 1735296,
        "cost_usd": null,
        "fresh_input_tokens": 45352,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1780648,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1735296,
        "known_fresh_input_tokens": 45352,
        "known_input_tokens": 1780648,
        "known_output_tokens": 10783,
        "known_reasoning_output_tokens": 3439,
        "output_tokens": 10783,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3439,
        "reasoning_reported_output_tokens": 10783,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 53,
          "cache_write_input_tokens": 53,
          "cache_write_reported_input_tokens": 53,
          "cached_input_tokens": 53,
          "fresh_input_tokens": 53,
          "input_tokens": 53,
          "output_tokens": 53,
          "reasoning_output_tokens": 53,
          "reasoning_reported_output_tokens": 53
        },
        "request_attempts": 53,
        "response_count": 53,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1791431,
        "uncached_input_tokens": 45352,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11820,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11953,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12005,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12354,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 17277,
            "output_tokens": 203,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 21676,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 22004,
            "output_tokens": 130,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22530,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22938,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23487,
            "output_tokens": 445,
            "reasoning_output_tokens": 157
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24168,
            "output_tokens": 1007,
            "reasoning_output_tokens": 172
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25197,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25336,
            "output_tokens": 44,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25442,
            "output_tokens": 120,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25984,
            "output_tokens": 199,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 27844,
            "output_tokens": 937,
            "reasoning_output_tokens": 140
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28802,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29275,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29735,
            "output_tokens": 96,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29853,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30283,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30749,
            "output_tokens": 619,
            "reasoning_output_tokens": 461
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 32052,
            "output_tokens": 91,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 33317,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 34171,
            "output_tokens": 218,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34416,
            "output_tokens": 120,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 35402,
            "output_tokens": 301,
            "reasoning_output_tokens": 174
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36100,
            "output_tokens": 297,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36607,
            "output_tokens": 202,
            "reasoning_output_tokens": 112
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37201,
            "output_tokens": 58,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37321,
            "output_tokens": 415,
            "reasoning_output_tokens": 314
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38167,
            "output_tokens": 322,
            "reasoning_output_tokens": 233
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38879,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39316,
            "output_tokens": 81,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39776,
            "output_tokens": 119,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40286,
            "output_tokens": 198,
            "reasoning_output_tokens": 159
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40808,
            "output_tokens": 262,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41230,
            "output_tokens": 43,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41335,
            "output_tokens": 421,
            "reasoning_output_tokens": 292
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41824,
            "output_tokens": 249,
            "reasoning_output_tokens": 117
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42472,
            "output_tokens": 129,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42989,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43425,
            "output_tokens": 237,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 44063,
            "output_tokens": 452,
            "reasoning_output_tokens": 244
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44726,
            "output_tokens": 168,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45283,
            "output_tokens": 67,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45727,
            "output_tokens": 86,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 46189,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46664,
            "output_tokens": 120,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 46926,
            "output_tokens": 54,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47042,
            "output_tokens": 179,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 47709,
            "output_tokens": 779,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 48513,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 67,
        "model_tool_calls_by_name": {
          "read": 18,
          "ls": 1,
          "find": 4,
          "exec": 33,
          "devices": 2,
          "write": 3,
          "device_stop": 5,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 578 rosbridge control frames; 29 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 22
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/handover-microphone.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/workspace/.kinex/memos/handover-microphone.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-18-20261116/workspace/.kinex/tools/tcp_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 123,
        "observed_images": 24
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 29,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 232,
        "published_frames": 232,
        "decode_ok": true,
        "duration_s": 5.8,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f89ecb2d275269b636d1976068504333e3bc432ee60ef908b47f79b9a4752ba9",
        "source_sha256": {
          "third_person": "5aeb8b38cce883341cfb1f348f4e12b50b1be01bef5923d14d48531f35477396",
          "left_wrist": "74b2c6cf8e7ab5835dde18725b20dfd83e9f0e8a88ed559efbf9d8dc673d1112",
          "right_wrist": "b4d0d413471761927ea21c080559d13f3c940ff15fa00d55de0ad7c5568d98d3"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "d78437ddf6a2",
          "task": "robotwin/handover-mic-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 233,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 232,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23120000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 233,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 232,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23120000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 233,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 232,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23120000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-19-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/19",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "19",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 10.72,
        "step_limit": 400,
        "steps": 268,
        "success": true,
        "valid": true
      },
      "steps": 268,
      "simulation_time_s": 10.72,
      "wall_time_s": 812.688373,
      "timing": {
        "simulation_time_s": 10.72,
        "wall_time_s": 812.688373,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the medium sauce can and place it near the smooth gray pot with matte finish",
      "native_instruction": "Grab the medium sauce can and place it near the smooth gray pot with matte finish",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9325752960228295,
        "cache_read_input_tokens": 1181696,
        "cache_reported_input_tokens": 1267132,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1267132,
        "cached_input_tokens": 1181696,
        "cost_usd": null,
        "fresh_input_tokens": 85436,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1267132,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1181696,
        "known_fresh_input_tokens": 85436,
        "known_input_tokens": 1267132,
        "known_output_tokens": 6773,
        "known_reasoning_output_tokens": 1020,
        "output_tokens": 6773,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1020,
        "reasoning_reported_output_tokens": 6773,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 41,
          "cache_write_input_tokens": 41,
          "cache_write_reported_input_tokens": 41,
          "cached_input_tokens": 41,
          "fresh_input_tokens": 41,
          "input_tokens": 41,
          "output_tokens": 41,
          "reasoning_output_tokens": 41,
          "reasoning_reported_output_tokens": 41
        },
        "request_attempts": 41,
        "response_count": 41,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1273905,
        "uncached_input_tokens": 85436,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11816,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11856,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11989,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12037,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12330,
            "output_tokens": 223,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 17398,
            "output_tokens": 198,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 22366,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22694,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22885,
            "output_tokens": 213,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23494,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23902,
            "output_tokens": 176,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 25922,
            "output_tokens": 284,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26599,
            "output_tokens": 51,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26856,
            "output_tokens": 282,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 27767,
            "output_tokens": 1081,
            "reasoning_output_tokens": 246
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28870,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 29763,
            "output_tokens": 104,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30681,
            "output_tokens": 58,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30949,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31452,
            "output_tokens": 849,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32322,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 33038,
            "output_tokens": 86,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33151,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33935,
            "output_tokens": 89,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34719,
            "output_tokens": 147,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34893,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35633,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36632,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 37545,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37642,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38623,
            "output_tokens": 139,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 39507,
            "output_tokens": 130,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 40385,
            "output_tokens": 178,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40837,
            "output_tokens": 170,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 41753,
            "output_tokens": 309,
            "reasoning_output_tokens": 223
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42806,
            "output_tokens": 104,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43651,
            "output_tokens": 212,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44005,
            "output_tokens": 84,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44485,
            "output_tokens": 208,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 44720,
            "output_tokens": 479,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45224,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 50,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 19,
          "bash": 1,
          "exec": 18,
          "devices": 2,
          "device": 1,
          "write": 3,
          "edit": 4
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 268 rosbridge control frames; 15 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 11
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-19-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 94,
        "observed_images": 20
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 15,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 108,
        "published_frames": 108,
        "decode_ok": true,
        "duration_s": 2.7,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "55ae625a2abe44f11241d990143cef9d2f9b5c3a05b86a99514877d25906eafc",
        "source_sha256": {
          "third_person": "578df9b27dbd878f4a97ee51ae54443de22cd36e50ee3bf535cd03f63fcdd52c",
          "left_wrist": "15c9758cbd5539ba21deb993e2ad4590ef7945ad21546a8b5b911aaf939113ea",
          "right_wrist": "0f1c3763f400612e1f7dcc66f8e0bcfa1fd95ee1181190de30f794590d6087b8"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "d2365216acb7",
          "task": "robotwin/move-can-pot-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 109,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 108,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10720000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 109,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 108,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10720000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 109,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 108,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10720000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-20-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/20",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "20",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 13.36,
        "step_limit": 400,
        "steps": 334,
        "success": true,
        "valid": true
      },
      "steps": 334,
      "simulation_time_s": 13.36,
      "wall_time_s": 942.827999,
      "timing": {
        "simulation_time_s": 13.36,
        "wall_time_s": 942.827999,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the medium-sized bottle and put it onto the pad",
      "native_instruction": "Pick the medium-sized bottle and put it onto the pad",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9759002011316591,
        "cache_read_input_tokens": 1246976,
        "cache_reported_input_tokens": 1277770,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1277770,
        "cached_input_tokens": 1246976,
        "cost_usd": null,
        "fresh_input_tokens": 30794,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1277770,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1246976,
        "known_fresh_input_tokens": 30794,
        "known_input_tokens": 1277770,
        "known_output_tokens": 8657,
        "known_reasoning_output_tokens": 1534,
        "output_tokens": 8657,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1534,
        "reasoning_reported_output_tokens": 8657,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 41,
          "cache_write_input_tokens": 41,
          "cache_write_reported_input_tokens": 41,
          "cached_input_tokens": 41,
          "fresh_input_tokens": 41,
          "input_tokens": 41,
          "output_tokens": 41,
          "reasoning_output_tokens": 41,
          "reasoning_reported_output_tokens": 41
        },
        "request_attempts": 41,
        "response_count": 41,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1286427,
        "uncached_input_tokens": 30794,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11814,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11854,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11987,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12035,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12328,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16585,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 20401,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 23635,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 23963,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24153,
            "output_tokens": 68,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24590,
            "output_tokens": 298,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 26760,
            "output_tokens": 481,
            "reasoning_output_tokens": 157
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27476,
            "output_tokens": 920,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28418,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 29238,
            "output_tokens": 154,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 31053,
            "output_tokens": 60,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31175,
            "output_tokens": 329,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31890,
            "output_tokens": 418,
            "reasoning_output_tokens": 192
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32547,
            "output_tokens": 857,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33425,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 34125,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34208,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34584,
            "output_tokens": 139,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35015,
            "output_tokens": 156,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35360,
            "output_tokens": 123,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35776,
            "output_tokens": 153,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36226,
            "output_tokens": 168,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36636,
            "output_tokens": 281,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37781,
            "output_tokens": 246,
            "reasoning_output_tokens": 167
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38320,
            "output_tokens": 60,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38402,
            "output_tokens": 328,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39120,
            "output_tokens": 340,
            "reasoning_output_tokens": 135
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39684,
            "output_tokens": 376,
            "reasoning_output_tokens": 117
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40671,
            "output_tokens": 407,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41443,
            "output_tokens": 124,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41863,
            "output_tokens": 141,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42146,
            "output_tokens": 53,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42261,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42454,
            "output_tokens": 425,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 42906,
            "output_tokens": 531,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43462,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 51,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "bash": 1,
          "exec": 21,
          "devices": 1,
          "device": 2,
          "write": 3,
          "device_stop": 3,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 334 rosbridge control frames; 16 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 10
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-20-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 95,
        "observed_images": 13
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 16,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 135,
        "published_frames": 135,
        "decode_ok": true,
        "duration_s": 3.375,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "ed93737f181712cd789ffd3e4cbddcbe652c2ed03e56dc6e08523278ff47a6eb",
        "source_sha256": {
          "third_person": "ea7da7bdcfa9ad860e1461ee61ae56ad600e24f2891d9787d76660b4d079691c",
          "left_wrist": "95b358b010cb1c7335ac79b225b283604bbab01b8771c5638d8cce36a06627f1",
          "right_wrist": "16d441cb37d937a62b8a5cac2551dd1b55ddf3407de856b6c8560542f366c629"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "7511d0b15f0e",
          "task": "robotwin/move-pillbottle-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 135,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 135,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13360000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 135,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 135,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13360000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 135,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 135,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13360000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-21-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/21",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "21",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 10.6,
        "step_limit": 400,
        "steps": 265,
        "success": false,
        "valid": true
      },
      "steps": 265,
      "simulation_time_s": 10.6,
      "wall_time_s": 737.269102,
      "timing": {
        "simulation_time_s": 10.6,
        "wall_time_s": 737.269102,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the small packet of playingcards and move it outward.",
      "native_instruction": "Pick up the small packet of playingcards and move it outward.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9031688395938263,
        "cache_read_input_tokens": 801920,
        "cache_reported_input_tokens": 887896,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 887896,
        "cached_input_tokens": 801920,
        "cost_usd": null,
        "fresh_input_tokens": 85976,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 887896,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 801920,
        "known_fresh_input_tokens": 85976,
        "known_input_tokens": 887896,
        "known_output_tokens": 5246,
        "known_reasoning_output_tokens": 853,
        "output_tokens": 5246,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 853,
        "reasoning_reported_output_tokens": 5246,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 29,
          "cache_write_input_tokens": 29,
          "cache_write_reported_input_tokens": 29,
          "cached_input_tokens": 29,
          "fresh_input_tokens": 29,
          "input_tokens": 29,
          "output_tokens": 29,
          "reasoning_output_tokens": 29,
          "reasoning_reported_output_tokens": 29
        },
        "request_attempts": 29,
        "response_count": 29,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 893142,
        "uncached_input_tokens": 85976,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11991,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 16002,
            "output_tokens": 43,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 16255,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16256,
            "input_tokens": 20512,
            "output_tokens": 206,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 24361,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24689,
            "output_tokens": 154,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25238,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25646,
            "output_tokens": 155,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 27534,
            "output_tokens": 251,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28137,
            "output_tokens": 77,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28680,
            "output_tokens": 233,
            "reasoning_output_tokens": 191
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 29952,
            "output_tokens": 757,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30731,
            "output_tokens": 144,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 35849,
            "output_tokens": 69,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35980,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36232,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 37261,
            "output_tokens": 820,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 38102,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38831,
            "output_tokens": 128,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39598,
            "output_tokens": 121,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 39741,
            "output_tokens": 176,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 40310,
            "output_tokens": 132,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 41077,
            "output_tokens": 266,
            "reasoning_output_tokens": 182
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41986,
            "output_tokens": 220,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42882,
            "output_tokens": 123,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 43033,
            "output_tokens": 555,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43610,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 37,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 17,
          "find": 1,
          "bash": 1,
          "devices": 1,
          "exec": 10,
          "write": 3,
          "device_stop": 2,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 265 rosbridge control frames; 25 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 5
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/playingcards.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/workspace/.kinex/memos/playingcards.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-21-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 69,
        "observed_images": 9
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 25,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 107,
        "published_frames": 107,
        "decode_ok": true,
        "duration_s": 2.675,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c4435e34097b51704f5cc176946710ab9952e067968628991f8c1a1bb4589055",
        "source_sha256": {
          "third_person": "de9712e31b709a74d8f469896b8a7f4326575b8a20a612666d10587a65d52f9a",
          "left_wrist": "50d27c37d111bac0333a16d1e0c76754e6f0d00a7650c970ec2dfb517890a6ce",
          "right_wrist": "549e39d01422d724cba0358d9c126d21b508dd4803705e9d29710c97f4f2f1d3"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ef5d88b5b95e",
          "task": "robotwin/move-playingcard-away-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 108,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 107,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 108,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 107,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 108,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 107,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-22-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/22",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "22",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 13.0,
        "step_limit": 400,
        "steps": 325,
        "success": false,
        "valid": true
      },
      "steps": 325,
      "simulation_time_s": 13.0,
      "wall_time_s": 837.507922,
      "timing": {
        "simulation_time_s": 13.0,
        "wall_time_s": 837.507922,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the red plastic stapler and set it on Red mat",
      "native_instruction": "Grab the red plastic stapler and set it on Red mat",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9646802436203725,
        "cache_read_input_tokens": 1039360,
        "cache_reported_input_tokens": 1077414,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1077414,
        "cached_input_tokens": 1039360,
        "cost_usd": null,
        "fresh_input_tokens": 38054,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1077414,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1039360,
        "known_fresh_input_tokens": 38054,
        "known_input_tokens": 1077414,
        "known_output_tokens": 7623,
        "known_reasoning_output_tokens": 1723,
        "output_tokens": 7623,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1723,
        "reasoning_reported_output_tokens": 7623,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 36,
          "cache_write_input_tokens": 36,
          "cache_write_reported_input_tokens": 36,
          "cached_input_tokens": 36,
          "fresh_input_tokens": 36,
          "input_tokens": 36,
          "output_tokens": 36,
          "reasoning_output_tokens": 36,
          "reasoning_reported_output_tokens": 36
        },
        "request_attempts": 36,
        "response_count": 36,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1085037,
        "uncached_input_tokens": 38054,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11816,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11856,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11989,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14465,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 18722,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 22538,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22866,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23057,
            "output_tokens": 132,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 24526,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24934,
            "output_tokens": 286,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25917,
            "output_tokens": 149,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26963,
            "output_tokens": 294,
            "reasoning_output_tokens": 252
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 28296,
            "output_tokens": 790,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29108,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29272,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30424,
            "output_tokens": 52,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30538,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30790,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 32009,
            "output_tokens": 1074,
            "reasoning_output_tokens": 198
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33104,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33576,
            "output_tokens": 374,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34127,
            "output_tokens": 199,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34700,
            "output_tokens": 43,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34799,
            "output_tokens": 122,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35163,
            "output_tokens": 125,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35754,
            "output_tokens": 183,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36160,
            "output_tokens": 133,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36668,
            "output_tokens": 373,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37192,
            "output_tokens": 197,
            "reasoning_output_tokens": 106
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37762,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38222,
            "output_tokens": 176,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38620,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39082,
            "output_tokens": 993,
            "reasoning_output_tokens": 600
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40437,
            "output_tokens": 261,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40726,
            "output_tokens": 485,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41236,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 44,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 17,
          "find": 1,
          "devices": 1,
          "device": 2,
          "exec": 17,
          "write": 3,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 325 rosbridge control frames; 20 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 8
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-22-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 83,
        "observed_images": 17
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 20,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 131,
        "published_frames": 131,
        "decode_ok": true,
        "duration_s": 3.275,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "0267e508dcdf6c368522f2d9205d29d14547953217438b2becba2a0202ca007e",
        "source_sha256": {
          "third_person": "7a7f442a4797e3b803f75f3a3303b884aeff9a909c748beff83aaf0643ee6f97",
          "left_wrist": "e2bcd84e6caac9052a0451578b23b97f6e383e98d612720f8dca2438b8633ecc",
          "right_wrist": "c1008fc40a7ff882172881f3f4ab06a7dbebee6185726135c1a79140d6fd4106"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "d007f5fa166f",
          "task": "robotwin/move-stapler-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 132,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 131,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13000000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 132,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 131,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13000000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 132,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 131,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13000000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-23-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/23",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "23",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 25.16,
        "step_limit": 700,
        "steps": 629,
        "success": false,
        "valid": true
      },
      "steps": 629,
      "simulation_time_s": 25.16,
      "wall_time_s": 1426.755763,
      "timing": {
        "simulation_time_s": 25.16,
        "wall_time_s": 1426.755763,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift the black laptop using the right arm.",
      "native_instruction": "Lift the black laptop using the right arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9599607987259586,
        "cache_read_input_tokens": 1959040,
        "cache_reported_input_tokens": 2040750,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2040750,
        "cached_input_tokens": 1959040,
        "cost_usd": null,
        "fresh_input_tokens": 81710,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2040750,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1959040,
        "known_fresh_input_tokens": 81710,
        "known_input_tokens": 2040750,
        "known_output_tokens": 13349,
        "known_reasoning_output_tokens": 4822,
        "output_tokens": 13349,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 4822,
        "reasoning_reported_output_tokens": 13349,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 56,
          "cache_write_input_tokens": 56,
          "cache_write_reported_input_tokens": 56,
          "cached_input_tokens": 56,
          "fresh_input_tokens": 56,
          "input_tokens": 56,
          "output_tokens": 56,
          "reasoning_output_tokens": 56,
          "reasoning_reported_output_tokens": 56
        },
        "request_attempts": 56,
        "response_count": 56,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2054099,
        "uncached_input_tokens": 81710,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11810,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11850,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11983,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14348,
            "output_tokens": 216,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 17913,
            "output_tokens": 204,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 22510,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22838,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 24237,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24645,
            "output_tokens": 123,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25190,
            "output_tokens": 384,
            "reasoning_output_tokens": 121
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 26024,
            "output_tokens": 263,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 27097,
            "output_tokens": 210,
            "reasoning_output_tokens": 171
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 27773,
            "output_tokens": 1007,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 28802,
            "output_tokens": 222,
            "reasoning_output_tokens": 112
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29090,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 30029,
            "output_tokens": 73,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 30164,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30416,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31445,
            "output_tokens": 908,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 32374,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32974,
            "output_tokens": 318,
            "reasoning_output_tokens": 106
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33735,
            "output_tokens": 54,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33811,
            "output_tokens": 324,
            "reasoning_output_tokens": 228
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34642,
            "output_tokens": 625,
            "reasoning_output_tokens": 477
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35454,
            "output_tokens": 113,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36071,
            "output_tokens": 574,
            "reasoning_output_tokens": 303
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37075,
            "output_tokens": 205,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37785,
            "output_tokens": 86,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 37933,
            "output_tokens": 124,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 38491,
            "output_tokens": 87,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39273,
            "output_tokens": 61,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 39383,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39660,
            "output_tokens": 442,
            "reasoning_output_tokens": 281
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 40637,
            "output_tokens": 171,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41568,
            "output_tokens": 332,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42148,
            "output_tokens": 682,
            "reasoning_output_tokens": 341
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 42857,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43484,
            "output_tokens": 174,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 44159,
            "output_tokens": 89,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44275,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44859,
            "output_tokens": 51,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 44932,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45343,
            "output_tokens": 297,
            "reasoning_output_tokens": 181
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 46483,
            "output_tokens": 106,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 46651,
            "output_tokens": 455,
            "reasoning_output_tokens": 174
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47381,
            "output_tokens": 760,
            "reasoning_output_tokens": 481
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48000,
            "input_tokens": 48419,
            "output_tokens": 140,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 48768,
            "output_tokens": 271,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49397,
            "output_tokens": 315,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 50100,
            "output_tokens": 500,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50432,
            "input_tokens": 50990,
            "output_tokens": 38,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50816,
            "input_tokens": 51090,
            "output_tokens": 81,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 51539,
            "output_tokens": 845,
            "reasoning_output_tokens": 500
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50944,
            "input_tokens": 52683,
            "output_tokens": 97,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52736,
            "input_tokens": 52842,
            "output_tokens": 457,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53248,
            "input_tokens": 53320,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 67,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 17,
          "find": 1,
          "devices": 5,
          "exec": 26,
          "write": 3,
          "device_stop": 7,
          "device": 2,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 629 rosbridge control frames; 38 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 11
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/laptop.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/workspace/.kinex/memos/laptop.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-23-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 126,
        "observed_images": 20
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 38,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 253,
        "published_frames": 253,
        "decode_ok": true,
        "duration_s": 6.325,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "3efef32ce6be84dbe52e70f332d12391430b761b768979b503469e14a48607b1",
        "source_sha256": {
          "third_person": "7408a66466a6f2a0445ac117ce6fc5d1c5c5f71e84deda0c8899919f41b7c869",
          "left_wrist": "af55c1d1d6c4432ce039881c48b01672fc8235e694c6b3e4d88afe478b49d007",
          "right_wrist": "53258150469968d572f5d2ae4ed214711308f31246976444056140ab9f3d21f0"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "0d85e83190ad",
          "task": "robotwin/open-laptop-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 253,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 253,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 25160000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 253,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 253,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 25160000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 253,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 253,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 25160000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-24-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/24",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "24",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 11.040000000000001,
        "step_limit": 1500,
        "steps": 276,
        "success": true,
        "valid": true
      },
      "steps": 276,
      "simulation_time_s": 11.040000000000001,
      "wall_time_s": 621.994763,
      "timing": {
        "simulation_time_s": 11.040000000000001,
        "wall_time_s": 621.994763,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to pull the handle of the kitchen microwave with slanted shape.",
      "native_instruction": "Use the left arm to pull the handle of the kitchen microwave with slanted shape.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9318896282454676,
        "cache_read_input_tokens": 657792,
        "cache_reported_input_tokens": 705869,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 705869,
        "cached_input_tokens": 657792,
        "cost_usd": null,
        "fresh_input_tokens": 48077,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 705869,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 657792,
        "known_fresh_input_tokens": 48077,
        "known_input_tokens": 705869,
        "known_output_tokens": 4616,
        "known_reasoning_output_tokens": 774,
        "output_tokens": 4616,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 774,
        "reasoning_reported_output_tokens": 4616,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 27,
          "cache_write_input_tokens": 27,
          "cache_write_reported_input_tokens": 27,
          "cached_input_tokens": 27,
          "fresh_input_tokens": 27,
          "input_tokens": 27,
          "output_tokens": 27,
          "reasoning_output_tokens": 27,
          "reasoning_reported_output_tokens": 27
        },
        "request_attempts": 27,
        "response_count": 27,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 710485,
        "uncached_input_tokens": 48077,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11952,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 12000,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12293,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12497,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 16754,
            "output_tokens": 198,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 20595,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 21065,
            "output_tokens": 70,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16768,
            "input_tokens": 21945,
            "output_tokens": 148,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 23874,
            "output_tokens": 428,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25921,
            "output_tokens": 167,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 28718,
            "output_tokens": 1097,
            "reasoning_output_tokens": 169
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29836,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30402,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 30459,
            "output_tokens": 136,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30622,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 31134,
            "output_tokens": 125,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31693,
            "output_tokens": 69,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31784,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32147,
            "output_tokens": 247,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32692,
            "output_tokens": 195,
            "reasoning_output_tokens": 112
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33320,
            "output_tokens": 290,
            "reasoning_output_tokens": 178
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34046,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34273,
            "output_tokens": 49,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34344,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34537,
            "output_tokens": 587,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35147,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 40,
        "model_tool_calls_by_name": {
          "read": 18,
          "ls": 1,
          "bash": 2,
          "devices": 2,
          "exec": 9,
          "device": 2,
          "write": 2,
          "device_stop": 3,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 276 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 7
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/open-microwave.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/workspace/.kinex/memos/open-microwave.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-24-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 70,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1500,
        "max_control_steps": 1500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 112,
        "published_frames": 112,
        "decode_ok": true,
        "duration_s": 2.8,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "78346ac826b5260409eea7a06926e8381b08a50bf9fdd9ab53d787170b7208da",
        "source_sha256": {
          "third_person": "d4a4721b7d75032ab8f35e6f119e4ae401774234c0cf9c6a2d3ed8c52e0cb367",
          "left_wrist": "5edce83e7f72454362a95c6f26dea730e49340a4fd67f8c5e779a2fa853b3712",
          "right_wrist": "f698d72ea23202c126ac2af51020ee60f764beb89f4f47c97207f3e0e74743e4"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "5d36697faeab",
          "task": "robotwin/open-microwave-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 112,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 112,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11040000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 112,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 112,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11040000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 112,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 112,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11040000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": false
    },
    {
      "id": "onboard-kinex-tcp-posestate-25-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/25",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "25",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 2
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "step_limit",
        "simulation_time_s": 16.0,
        "step_limit": 400,
        "steps": 400,
        "success": false,
        "valid": true
      },
      "steps": 400,
      "simulation_time_s": 16.0,
      "wall_time_s": 1105.141466,
      "timing": {
        "simulation_time_s": 16.0,
        "wall_time_s": 1105.141466,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift the bottle with attached loop on cap and the plastic bottle with blue label simultaneously with separate arms.",
      "native_instruction": "Lift the bottle with attached loop on cap and the plastic bottle with blue label simultaneously with separate arms.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9631532246765984,
        "cache_read_input_tokens": 1484928,
        "cache_reported_input_tokens": 1541736,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1541736,
        "cached_input_tokens": 1484928,
        "cost_usd": null,
        "fresh_input_tokens": 56808,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1541736,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1484928,
        "known_fresh_input_tokens": 56808,
        "known_input_tokens": 1541736,
        "known_output_tokens": 9208,
        "known_reasoning_output_tokens": 2566,
        "output_tokens": 9208,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2566,
        "reasoning_reported_output_tokens": 9208,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 45,
          "cache_write_input_tokens": 45,
          "cache_write_reported_input_tokens": 45,
          "cached_input_tokens": 45,
          "fresh_input_tokens": 45,
          "input_tokens": 45,
          "output_tokens": 45,
          "reasoning_output_tokens": 45,
          "reasoning_reported_output_tokens": 45
        },
        "request_attempts": 45,
        "response_count": 45,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1550944,
        "uncached_input_tokens": 56808,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11825,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11865,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11998,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14305,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14208,
            "input_tokens": 14507,
            "output_tokens": 223,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14592,
            "input_tokens": 19575,
            "output_tokens": 195,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 24214,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24542,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24732,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25140,
            "output_tokens": 310,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 27282,
            "output_tokens": 958,
            "reasoning_output_tokens": 121
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28262,
            "output_tokens": 110,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 29032,
            "output_tokens": 140,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29830,
            "output_tokens": 165,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 31656,
            "output_tokens": 950,
            "reasoning_output_tokens": 210
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32627,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33356,
            "output_tokens": 72,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33490,
            "output_tokens": 80,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 34008,
            "output_tokens": 230,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34506,
            "output_tokens": 65,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34633,
            "output_tokens": 349,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35753,
            "output_tokens": 565,
            "reasoning_output_tokens": 346
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36660,
            "output_tokens": 115,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37208,
            "output_tokens": 96,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37535,
            "output_tokens": 316,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38141,
            "output_tokens": 462,
            "reasoning_output_tokens": 212
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38630,
            "output_tokens": 358,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39243,
            "output_tokens": 191,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 40126,
            "output_tokens": 431,
            "reasoning_output_tokens": 360
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40995,
            "output_tokens": 72,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41129,
            "output_tokens": 48,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41228,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41421,
            "output_tokens": 240,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41824,
            "output_tokens": 111,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42630,
            "output_tokens": 342,
            "reasoning_output_tokens": 222
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 43633,
            "output_tokens": 290,
            "reasoning_output_tokens": 130
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 44044,
            "output_tokens": 130,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44192,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44953,
            "output_tokens": 158,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45605,
            "output_tokens": 135,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 46183,
            "output_tokens": 130,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46418,
            "output_tokens": 107,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 47189,
            "output_tokens": 86,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47556,
            "output_tokens": 475,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47872,
            "input_tokens": 48055,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 56,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 17,
          "find": 1,
          "bash": 1,
          "exec": 21,
          "devices": 4,
          "device": 4,
          "write": 3,
          "device_stop": 3,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 400 rosbridge control frames; 48 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 3,
          "direct_motion_exec_calls": 11
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/bottle_manipulation.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/workspace/.kinex/memos/bottle_manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-25-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 104,
        "observed_images": 22
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 48,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 161,
        "published_frames": 161,
        "decode_ok": true,
        "duration_s": 4.025,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "53768c12849df94a28fc127c20d4773f8f916d9963ee5e474c967cb6d0a7f256",
        "source_sha256": {
          "third_person": "1e6a734debff6ec78c781aa2cc8574fb8452087ab22c147c00d95a26cd5ec776",
          "left_wrist": "16dbb94ba07c0cc97ec19ca81f0468ee0e37aee4202bdcba16da1f47daacdcce",
          "right_wrist": "1e95c76d347f53b749612f37cb065b7ae6cfe9f4f1edd7383e938cfcd1ddb3dd"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ab3622e16852",
          "task": "robotwin/pick-diverse-bottles-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 162,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16000000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 162,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16000000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 162,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16000000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-26-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/26",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "26",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 11.6,
        "step_limit": 400,
        "steps": 290,
        "success": false,
        "valid": true
      },
      "steps": 290,
      "simulation_time_s": 11.6,
      "wall_time_s": 720.928156,
      "timing": {
        "simulation_time_s": 11.6,
        "wall_time_s": 720.928156,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use one arm for the bottle with red screw cap and the other for the green bottle.",
      "native_instruction": "Use one arm for the bottle with red screw cap and the other for the green bottle.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9583663687796198,
        "cache_read_input_tokens": 534272,
        "cache_reported_input_tokens": 557482,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 557482,
        "cached_input_tokens": 534272,
        "cost_usd": null,
        "fresh_input_tokens": 23210,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 557482,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 534272,
        "known_fresh_input_tokens": 23210,
        "known_input_tokens": 557482,
        "known_output_tokens": 4801,
        "known_reasoning_output_tokens": 790,
        "output_tokens": 4801,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 790,
        "reasoning_reported_output_tokens": 4801,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 23,
          "cache_write_input_tokens": 23,
          "cache_write_reported_input_tokens": 23,
          "cached_input_tokens": 23,
          "fresh_input_tokens": 23,
          "input_tokens": 23,
          "output_tokens": 23,
          "reasoning_output_tokens": 23,
          "reasoning_reported_output_tokens": 23
        },
        "request_attempts": 23,
        "response_count": 23,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 562283,
        "uncached_input_tokens": 23210,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11822,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11862,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11995,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12045,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13380,
            "output_tokens": 202,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13440,
            "input_tokens": 15735,
            "output_tokens": 181,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 19967,
            "output_tokens": 192,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 24071,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24399,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24589,
            "output_tokens": 69,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25027,
            "output_tokens": 270,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25951,
            "output_tokens": 1070,
            "reasoning_output_tokens": 234
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27045,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27770,
            "output_tokens": 150,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28532,
            "output_tokens": 69,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28811,
            "output_tokens": 262,
            "reasoning_output_tokens": 220
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 30252,
            "output_tokens": 794,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31069,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31554,
            "output_tokens": 89,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32023,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32514,
            "output_tokens": 343,
            "reasoning_output_tokens": 104
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33246,
            "output_tokens": 551,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33823,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 35,
        "model_tool_calls_by_name": {
          "ls": 5,
          "read": 16,
          "find": 1,
          "exec": 8,
          "devices": 1,
          "device": 1,
          "write": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 290 rosbridge control frames; 45 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 3
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/dual-bottles.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/workspace/.kinex/memos/dual-bottles.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-26-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 61,
        "observed_images": 12
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 45,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 117,
        "published_frames": 117,
        "decode_ok": true,
        "duration_s": 2.925,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "08afdc5c4d21b40575dfa722a74960ff4a1993d42f8a013c786b1396c973d2e2",
        "source_sha256": {
          "third_person": "33f512c7b90ba9708fa49bde6c581ed26282040974c188f3078de7fa1d38b4a1",
          "left_wrist": "d13871c2b7f8ffde2da7e8e098dafb2bcb57d0b92d5ef8468d26ea8493216102",
          "right_wrist": "9e738fed14b2b31d6860b38178f67e81b1438748de18e9b25f5d6d45326059dc"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "0c441e31cccd",
          "task": "robotwin/pick-dual-bottles-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 118,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 117,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 118,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 117,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 118,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 117,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-27-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/27",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "27",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 10.16,
        "step_limit": 400,
        "steps": 254,
        "success": true,
        "valid": true
      },
      "steps": 254,
      "simulation_time_s": 10.16,
      "wall_time_s": 690.842356,
      "timing": {
        "simulation_time_s": 10.16,
        "wall_time_s": 690.842356,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Position the coffee-box with mug image on the left of the mouse with scroll wheel",
      "native_instruction": "Position the coffee-box with mug image on the left of the mouse with scroll wheel",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9521647278162974,
        "cache_read_input_tokens": 863104,
        "cache_reported_input_tokens": 906465,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 906465,
        "cached_input_tokens": 863104,
        "cost_usd": null,
        "fresh_input_tokens": 43361,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 906465,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 863104,
        "known_fresh_input_tokens": 43361,
        "known_input_tokens": 906465,
        "known_output_tokens": 5246,
        "known_reasoning_output_tokens": 650,
        "output_tokens": 5246,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 650,
        "reasoning_reported_output_tokens": 5246,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 31,
          "cache_write_input_tokens": 31,
          "cache_write_reported_input_tokens": 31,
          "cached_input_tokens": 31,
          "fresh_input_tokens": 31,
          "input_tokens": 31,
          "output_tokens": 31,
          "reasoning_output_tokens": 31,
          "reasoning_reported_output_tokens": 31
        },
        "request_attempts": 31,
        "response_count": 31,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 911711,
        "uncached_input_tokens": 43361,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11991,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12335,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 15569,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15488,
            "input_tokens": 15728,
            "output_tokens": 171,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 18873,
            "output_tokens": 196,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 23801,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24129,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24319,
            "output_tokens": 84,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25442,
            "output_tokens": 318,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26155,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26563,
            "output_tokens": 317,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 28719,
            "output_tokens": 926,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29668,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 33466,
            "output_tokens": 60,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33588,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33840,
            "output_tokens": 100,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 35119,
            "output_tokens": 101,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35686,
            "output_tokens": 888,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36595,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37114,
            "output_tokens": 145,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37673,
            "output_tokens": 153,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38046,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38530,
            "output_tokens": 193,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39132,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39620,
            "output_tokens": 171,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 39933,
            "output_tokens": 57,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40052,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40244,
            "output_tokens": 590,
            "reasoning_output_tokens": 131
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 40859,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 39,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 17,
          "bash": 2,
          "exec": 11,
          "devices": 1,
          "device": 2,
          "write": 3,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 254 rosbridge control frames; 15 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 7
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin_manipulation.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/workspace/.kinex/memos/robotwin_manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-27-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 73,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 15,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 103,
        "published_frames": 103,
        "decode_ok": true,
        "duration_s": 2.575,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "9418bf165d092c96378c4ca1f90037c8887b24f440f4e5d951cd7b0fe058b9f4",
        "source_sha256": {
          "third_person": "2fc588f31f2cc52564c602ea103e53beda628d164d46ba3eb63277e702d7343e",
          "left_wrist": "ee66c8d9673eeaaf9f457b92acd899bc99710570993a70f81527756cde536ea5",
          "right_wrist": "ea92404854403c292a330a9d3832ada9db4efaebf25dfec2316e4e7dcfde2f66"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "f6c4306d70b6",
          "task": "robotwin/place-a2b-left-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 103,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 103,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10160000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 103,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 103,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10160000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 103,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 103,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10160000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-28-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/28",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "28",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.96,
        "step_limit": 400,
        "steps": 399,
        "success": false,
        "valid": true
      },
      "steps": 399,
      "simulation_time_s": 15.96,
      "wall_time_s": 1121.898895,
      "timing": {
        "simulation_time_s": 15.96,
        "wall_time_s": 1121.898895,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Put the tiny pink car with four wheels to the right of the red, blue, green, yellow rubikscube",
      "native_instruction": "Put the tiny pink car with four wheels to the right of the red, blue, green, yellow rubikscube",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9710994538439011,
        "cache_read_input_tokens": 1512064,
        "cache_reported_input_tokens": 1557064,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1557064,
        "cached_input_tokens": 1512064,
        "cost_usd": null,
        "fresh_input_tokens": 45000,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1557064,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1512064,
        "known_fresh_input_tokens": 45000,
        "known_input_tokens": 1557064,
        "known_output_tokens": 9742,
        "known_reasoning_output_tokens": 4122,
        "output_tokens": 9742,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 4122,
        "reasoning_reported_output_tokens": 9742,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 46,
          "cache_write_input_tokens": 46,
          "cache_write_reported_input_tokens": 46,
          "cached_input_tokens": 46,
          "fresh_input_tokens": 46,
          "input_tokens": 46,
          "output_tokens": 46,
          "reasoning_output_tokens": 46,
          "reasoning_reported_output_tokens": 46
        },
        "request_attempts": 46,
        "response_count": 46,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1566806,
        "uncached_input_tokens": 45000,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11826,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11866,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11999,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12343,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18233,
            "output_tokens": 193,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 22404,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22732,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 22924,
            "output_tokens": 291,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 26235,
            "output_tokens": 55,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26611,
            "output_tokens": 268,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27156,
            "output_tokens": 961,
            "reasoning_output_tokens": 183
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28139,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28275,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 29883,
            "output_tokens": 90,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30183,
            "output_tokens": 53,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30298,
            "output_tokens": 741,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31060,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31768,
            "output_tokens": 56,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 33003,
            "output_tokens": 71,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33101,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33581,
            "output_tokens": 137,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34111,
            "output_tokens": 143,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34336,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 35115,
            "output_tokens": 61,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 36047,
            "output_tokens": 108,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36548,
            "output_tokens": 308,
            "reasoning_output_tokens": 240
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36975,
            "output_tokens": 224,
            "reasoning_output_tokens": 143
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37595,
            "output_tokens": 275,
            "reasoning_output_tokens": 187
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38264,
            "output_tokens": 482,
            "reasoning_output_tokens": 376
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 38814,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39300,
            "output_tokens": 281,
            "reasoning_output_tokens": 195
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39975,
            "output_tokens": 779,
            "reasoning_output_tokens": 691
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41147,
            "output_tokens": 180,
            "reasoning_output_tokens": 100
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41717,
            "output_tokens": 61,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41833,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42025,
            "output_tokens": 205,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 43128,
            "output_tokens": 451,
            "reasoning_output_tokens": 345
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43778,
            "output_tokens": 241,
            "reasoning_output_tokens": 152
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44410,
            "output_tokens": 312,
            "reasoning_output_tokens": 223
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45115,
            "output_tokens": 611,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 46118,
            "output_tokens": 160,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46671,
            "output_tokens": 129,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 46882,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47365,
            "output_tokens": 637,
            "reasoning_output_tokens": 138
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47872,
            "input_tokens": 48027,
            "output_tokens": 59,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48000,
            "input_tokens": 48148,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 57,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 17,
          "bash": 1,
          "exec": 27,
          "devices": 2,
          "device": 3,
          "write": 3,
          "device_stop": 2,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 399 rosbridge control frames; 18 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 18
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin_manipulation.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/workspace/.kinex/memos/robotwin_manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-28-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 106,
        "observed_images": 36
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 18,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 161,
        "published_frames": 161,
        "decode_ok": true,
        "duration_s": 4.025,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "b160d03266f4352c8fa7b2ead82432f8c4c254375070a9ac9e33bb85181319c3",
        "source_sha256": {
          "third_person": "1619cfa6491b1fdd1a6a5e25a93da9c19e12f3ea0e753768bb1d84843a9d24f1",
          "left_wrist": "1b3955ed1765041df142b3ee6aaf6761f1da1e7fc2513bf96700fdc7a672c4d1",
          "right_wrist": "cfcd99c14b7339fa3e01c7eab5199efbce03df3736bab49bc424aefad7767781"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "5c86e572b028",
          "task": "robotwin/place-a2b-right-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-29-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/29",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "29",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 18.52,
        "step_limit": 700,
        "steps": 463,
        "success": true,
        "valid": true
      },
      "steps": 463,
      "simulation_time_s": 18.52,
      "wall_time_s": 941.854759,
      "timing": {
        "simulation_time_s": 18.52,
        "wall_time_s": 941.854759,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move both the soft bread with golden edges and the soft rectangular bread loaf inside the white basket with beige bottom.",
      "native_instruction": "Move both the soft bread with golden edges and the soft rectangular bread loaf inside the white basket with beige bottom.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9673439096287596,
        "cache_read_input_tokens": 1024512,
        "cache_reported_input_tokens": 1059098,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1059098,
        "cached_input_tokens": 1024512,
        "cost_usd": null,
        "fresh_input_tokens": 34586,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1059098,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1024512,
        "known_fresh_input_tokens": 34586,
        "known_input_tokens": 1059098,
        "known_output_tokens": 6147,
        "known_reasoning_output_tokens": 1128,
        "output_tokens": 6147,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1128,
        "reasoning_reported_output_tokens": 6147,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1065245,
        "uncached_input_tokens": 34586,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11823,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11863,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11996,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 15662,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15616,
            "input_tokens": 15955,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16128,
            "input_tokens": 21845,
            "output_tokens": 187,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 25684,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26012,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26202,
            "output_tokens": 161,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 29536,
            "output_tokens": 318,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30379,
            "output_tokens": 73,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30918,
            "output_tokens": 983,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31924,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32750,
            "output_tokens": 173,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33617,
            "output_tokens": 48,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33727,
            "output_tokens": 115,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 35039,
            "output_tokens": 881,
            "reasoning_output_tokens": 131
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 35941,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36432,
            "output_tokens": 277,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37100,
            "output_tokens": 229,
            "reasoning_output_tokens": 118
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37694,
            "output_tokens": 65,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38104,
            "output_tokens": 111,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38580,
            "output_tokens": 301,
            "reasoning_output_tokens": 219
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39235,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39712,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40113,
            "output_tokens": 199,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40677,
            "output_tokens": 179,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41210,
            "output_tokens": 89,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41321,
            "output_tokens": 78,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41541,
            "output_tokens": 55,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41658,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42092,
            "output_tokens": 638,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42756,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 45,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 17,
          "find": 1,
          "bash": 1,
          "exec": 16,
          "devices": 2,
          "device": 1,
          "write": 3,
          "device_stop": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 463 rosbridge control frames; 37 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 10
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin_bread_basket.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/workspace/.kinex/memos/robotwin_bread_basket.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-29-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 81,
        "observed_images": 18
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 37,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 186,
        "published_frames": 186,
        "decode_ok": true,
        "duration_s": 4.65,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "fee73487c3c0b962e9c554d36bedcf48dd87e53c48093adaa271dfaf38e15370",
        "source_sha256": {
          "third_person": "fc63904187f654f0056acdb965b3a4e7cd7c24461466090c687085d6e1989dd0",
          "left_wrist": "5cce14dc797247e5a8974174ebe2e842215b6e070bd3ce7e0a9945792ea83a36",
          "right_wrist": "810d5e9e517b180393ee1d44037d249ffbcf58dd643328ccf518e83aeb813997"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "b16c4660ee92",
          "task": "robotwin/place-bread-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 187,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 186,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18520000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 187,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 186,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18520000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 187,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 186,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18520000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-30-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/30",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "30",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 7.84,
        "step_limit": 500,
        "steps": 196,
        "success": true,
        "valid": true
      },
      "steps": 196,
      "simulation_time_s": 7.84,
      "wall_time_s": 425.659773,
      "timing": {
        "simulation_time_s": 7.84,
        "wall_time_s": 425.659773,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move the spongy golden brown bread from the table into the versatile skillet for frying and sauteing.",
      "native_instruction": "Move the spongy golden brown bread from the table into the versatile skillet for frying and sauteing.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9093882039246096,
        "cache_read_input_tokens": 572288,
        "cache_reported_input_tokens": 629311,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 629311,
        "cached_input_tokens": 572288,
        "cost_usd": null,
        "fresh_input_tokens": 57023,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 629311,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 572288,
        "known_fresh_input_tokens": 57023,
        "known_input_tokens": 629311,
        "known_output_tokens": 4662,
        "known_reasoning_output_tokens": 986,
        "output_tokens": 4662,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 986,
        "reasoning_reported_output_tokens": 4662,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 27,
          "cache_write_input_tokens": 27,
          "cache_write_reported_input_tokens": 27,
          "cached_input_tokens": 27,
          "fresh_input_tokens": 27,
          "input_tokens": 27,
          "output_tokens": 27,
          "reasoning_output_tokens": 27,
          "reasoning_reported_output_tokens": 27
        },
        "request_attempts": 27,
        "response_count": 27,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 633973,
        "uncached_input_tokens": 57023,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11822,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11862,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11995,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12339,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16596,
            "output_tokens": 132,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 18833,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19063,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 19254,
            "output_tokens": 179,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 21286,
            "output_tokens": 304,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 21985,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22393,
            "output_tokens": 306,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 23389,
            "output_tokens": 110,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 23965,
            "output_tokens": 924,
            "reasoning_output_tokens": 159
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 24910,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 25192,
            "output_tokens": 252,
            "reasoning_output_tokens": 174
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25744,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26132,
            "output_tokens": 166,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 26572,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26954,
            "output_tokens": 96,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27737,
            "output_tokens": 144,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28576,
            "output_tokens": 483,
            "reasoning_output_tokens": 163
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29794,
            "output_tokens": 189,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30125,
            "output_tokens": 58,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30245,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30438,
            "output_tokens": 126,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30808,
            "output_tokens": 469,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31302,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 35,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 15,
          "bash": 1,
          "devices": 2,
          "device": 2,
          "exec": 11,
          "write": 2,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 196 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 7
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-30-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 65,
        "observed_images": 6
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 80,
        "published_frames": 80,
        "decode_ok": true,
        "duration_s": 2.0,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "6aab2cd57c843a1e2eb0d21224ffc4bd64fbc42221220f00f4ff14c5eba482c9",
        "source_sha256": {
          "third_person": "211027e404b5c52f805557455ff4ed7c1cc6e6f752d61cb9511c4abbb9d0793b",
          "left_wrist": "562c87ebb8a50cd4b49d6fa0ec7ddd3e48a94a76f9f73c6f6b2114616ae992a6",
          "right_wrist": "e254f5762339842ffd0aaeec8541c39d202d714efd45b27727384af3aa820393"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "41fef4ecb280",
          "task": "robotwin/place-bread-skillet-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 80,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 80,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7840000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 80,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 80,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7840000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 80,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 80,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7840000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": false
    },
    {
      "id": "onboard-kinex-tcp-posestate-31-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/31",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "31",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 19.56,
        "step_limit": 500,
        "steps": 489,
        "success": true,
        "valid": true
      },
      "steps": 489,
      "simulation_time_s": 19.56,
      "wall_time_s": 1186.143924,
      "timing": {
        "simulation_time_s": 19.56,
        "wall_time_s": 1186.143924,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the compact hamburg box and the Crispy potato fries with red box, place them on the green tray.",
      "native_instruction": "Pick the compact hamburg box and the Crispy potato fries with red box, place them on the green tray.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9622515401968492,
        "cache_read_input_tokens": 1534720,
        "cache_reported_input_tokens": 1594926,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1594926,
        "cached_input_tokens": 1534720,
        "cost_usd": null,
        "fresh_input_tokens": 60206,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1594926,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1534720,
        "known_fresh_input_tokens": 60206,
        "known_input_tokens": 1594926,
        "known_output_tokens": 8801,
        "known_reasoning_output_tokens": 2305,
        "output_tokens": 8801,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2305,
        "reasoning_reported_output_tokens": 8801,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 45,
          "cache_write_input_tokens": 45,
          "cache_write_reported_input_tokens": 45,
          "cached_input_tokens": 45,
          "fresh_input_tokens": 45,
          "input_tokens": 45,
          "output_tokens": 45,
          "reasoning_output_tokens": 45,
          "reasoning_reported_output_tokens": 45
        },
        "request_attempts": 45,
        "response_count": 45,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1603727,
        "uncached_input_tokens": 60206,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11823,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11989,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12037,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12386,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 12533,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12672,
            "input_tokens": 18278,
            "output_tokens": 177,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 23707,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24035,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24227,
            "output_tokens": 188,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 27156,
            "output_tokens": 302,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 28330,
            "output_tokens": 96,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28892,
            "output_tokens": 287,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29848,
            "output_tokens": 1115,
            "reasoning_output_tokens": 178
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 30986,
            "output_tokens": 110,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 32864,
            "output_tokens": 312,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33569,
            "output_tokens": 264,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34226,
            "output_tokens": 62,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34494,
            "output_tokens": 250,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35315,
            "output_tokens": 191,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 37168,
            "output_tokens": 159,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 38220,
            "output_tokens": 230,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38479,
            "output_tokens": 117,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 40188,
            "output_tokens": 66,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40316,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40568,
            "output_tokens": 138,
            "reasoning_output_tokens": 96
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 41695,
            "output_tokens": 801,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42517,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42993,
            "output_tokens": 58,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43113,
            "output_tokens": 165,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43647,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 44070,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44546,
            "output_tokens": 357,
            "reasoning_output_tokens": 238
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45273,
            "output_tokens": 225,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45867,
            "output_tokens": 146,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45952,
            "input_tokens": 46386,
            "output_tokens": 543,
            "reasoning_output_tokens": 425
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47299,
            "output_tokens": 72,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47741,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 48165,
            "output_tokens": 300,
            "reasoning_output_tokens": 221
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 48836,
            "output_tokens": 107,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48768,
            "input_tokens": 49314,
            "output_tokens": 159,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49408,
            "input_tokens": 49615,
            "output_tokens": 62,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 49739,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49664,
            "input_tokens": 49931,
            "output_tokens": 731,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50560,
            "input_tokens": 50687,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 59,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 18,
          "find": 4,
          "bash": 1,
          "exec": 24,
          "devices": 1,
          "device": 2,
          "write": 3,
          "edit": 1,
          "device_stop": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 489 rosbridge control frames; 57 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 4,
          "direct_motion_exec_calls": 13
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_step.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/workspace/.kinex/tools/cartesian_step.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-31-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 107,
        "observed_images": 29
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 57,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 197,
        "published_frames": 197,
        "decode_ok": true,
        "duration_s": 4.925,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f608da0649fdd5f792cbda6489ab61b081163328e9979d3cfe9a495ac0ef195c",
        "source_sha256": {
          "third_person": "9fa6489968b2605ee4490b2464317f72ef97df3b5757d42bb76f56b7c8e4223b",
          "left_wrist": "c71275f0bca802ef0231b6d75d68cfa61b9e1ebb74a44d3931cbd4a41033be1d",
          "right_wrist": "e8e87b3bf3382c04b5a953a8ac200a0169349d959ebc60e191a414836984ad78"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "f25f56bfeec3",
          "task": "robotwin/place-burger-fries-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 197,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 197,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 197,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 197,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 197,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 197,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-32-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/32",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "32",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 27.400000000000002,
        "step_limit": 700,
        "steps": 685,
        "success": false,
        "valid": true
      },
      "steps": 685,
      "simulation_time_s": 27.400000000000002,
      "wall_time_s": 1572.221225,
      "timing": {
        "simulation_time_s": 27.400000000000002,
        "wall_time_s": 1572.221225,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to put the red can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "native_instruction": "Use the left arm to put the red can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9788789698684947,
        "cache_read_input_tokens": 1972864,
        "cache_reported_input_tokens": 2015432,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2015432,
        "cached_input_tokens": 1972864,
        "cost_usd": null,
        "fresh_input_tokens": 42568,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2015432,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1972864,
        "known_fresh_input_tokens": 42568,
        "known_input_tokens": 2015432,
        "known_output_tokens": 12789,
        "known_reasoning_output_tokens": 4903,
        "output_tokens": 12789,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 4903,
        "reasoning_reported_output_tokens": 12789,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 54,
          "cache_write_input_tokens": 54,
          "cache_write_reported_input_tokens": 54,
          "cached_input_tokens": 54,
          "fresh_input_tokens": 54,
          "input_tokens": 54,
          "output_tokens": 54,
          "reasoning_output_tokens": 54,
          "reasoning_reported_output_tokens": 54
        },
        "request_attempts": 54,
        "response_count": 54,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2028221,
        "uncached_input_tokens": 42568,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11832,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11872,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 12005,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12349,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18239,
            "output_tokens": 71,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 19273,
            "output_tokens": 114,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 21456,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 21784,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 21975,
            "output_tokens": 84,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22428,
            "output_tokens": 266,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 24488,
            "output_tokens": 301,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 26140,
            "output_tokens": 270,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26889,
            "output_tokens": 1180,
            "reasoning_output_tokens": 362
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28091,
            "output_tokens": 110,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28783,
            "output_tokens": 95,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29088,
            "output_tokens": 184,
            "reasoning_output_tokens": 142
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 30261,
            "output_tokens": 949,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31231,
            "output_tokens": 105,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 32035,
            "output_tokens": 87,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 34143,
            "output_tokens": 177,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34347,
            "output_tokens": 105,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 35345,
            "output_tokens": 160,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35895,
            "output_tokens": 157,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36454,
            "output_tokens": 76,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36907,
            "output_tokens": 201,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37999,
            "output_tokens": 160,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38563,
            "output_tokens": 703,
            "reasoning_output_tokens": 315
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39467,
            "output_tokens": 401,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40298,
            "output_tokens": 98,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40784,
            "output_tokens": 346,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 42482,
            "output_tokens": 134,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42940,
            "output_tokens": 280,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43628,
            "output_tokens": 401,
            "reasoning_output_tokens": 319
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44415,
            "output_tokens": 84,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44561,
            "output_tokens": 1106,
            "reasoning_output_tokens": 981
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 46065,
            "output_tokens": 210,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46663,
            "output_tokens": 72,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 46797,
            "output_tokens": 310,
            "reasoning_output_tokens": 218
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47498,
            "output_tokens": 407,
            "reasoning_output_tokens": 372
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 47967,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47872,
            "input_tokens": 48444,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 48909,
            "output_tokens": 263,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49510,
            "output_tokens": 194,
            "reasoning_output_tokens": 112
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 50091,
            "output_tokens": 278,
            "reasoning_output_tokens": 242
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50304,
            "input_tokens": 50431,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50432,
            "input_tokens": 50922,
            "output_tokens": 104,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50944,
            "input_tokens": 51351,
            "output_tokens": 219,
            "reasoning_output_tokens": 130
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51456,
            "input_tokens": 51956,
            "output_tokens": 103,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51968,
            "input_tokens": 52121,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52096,
            "input_tokens": 52557,
            "output_tokens": 362,
            "reasoning_output_tokens": 337
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52736,
            "input_tokens": 53088,
            "output_tokens": 144,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53120,
            "input_tokens": 53624,
            "output_tokens": 783,
            "reasoning_output_tokens": 149
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54272,
            "input_tokens": 54430,
            "output_tokens": 103,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54400,
            "input_tokens": 54561,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 62,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 16,
          "bash": 1,
          "exec": 29,
          "devices": 3,
          "device": 2,
          "write": 3,
          "edit": 2,
          "device_stop": 5
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 685 rosbridge control frames; 40 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 19
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/place_can_basket.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/workspace/.kinex/memos/place_can_basket.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-32-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 119,
        "observed_images": 27
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 40,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 275,
        "published_frames": 275,
        "decode_ok": true,
        "duration_s": 6.875,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "140da05c4c6c63836d25e0d2e985073118a8f45296ed24ff2127f39e8f5e1f64",
        "source_sha256": {
          "third_person": "07e5c01f9a09739218159144817d541bbf6227f1ac0449696917f0b91b2a0b6b",
          "left_wrist": "087e1e40ade095e15b9c4e218c8de2c7dc4cb292c9738dba5b946bc956cc42e3",
          "right_wrist": "caf96580eeefec43f0c2c2e835851ba25fedf84bbb52ca7c47fa67a0dac4fe45"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "a4eb728f6675",
          "task": "robotwin/place-can-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 276,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 275,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27400000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 276,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 275,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27400000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 276,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 275,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27400000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-33-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/33",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "33",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 18.56,
        "step_limit": 800,
        "steps": 464,
        "success": true,
        "valid": true
      },
      "steps": 464,
      "simulation_time_s": 18.56,
      "wall_time_s": 900.504197,
      "timing": {
        "simulation_time_s": 18.56,
        "wall_time_s": 900.504197,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the small red soda can and the lightweight cylindrical metal can in parallel, then put both into the rectangular plasticbox.",
      "native_instruction": "Grab the small red soda can and the lightweight cylindrical metal can in parallel, then put both into the rectangular plasticbox.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9187849292901958,
        "cache_read_input_tokens": 828416,
        "cache_reported_input_tokens": 901643,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 901643,
        "cached_input_tokens": 828416,
        "cost_usd": null,
        "fresh_input_tokens": 73227,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 901643,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 828416,
        "known_fresh_input_tokens": 73227,
        "known_input_tokens": 901643,
        "known_output_tokens": 6320,
        "known_reasoning_output_tokens": 1233,
        "output_tokens": 6320,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1233,
        "reasoning_reported_output_tokens": 6320,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 32,
          "cache_write_input_tokens": 32,
          "cache_write_reported_input_tokens": 32,
          "cached_input_tokens": 32,
          "fresh_input_tokens": 32,
          "input_tokens": 32,
          "output_tokens": 32,
          "reasoning_output_tokens": 32,
          "reasoning_reported_output_tokens": 32
        },
        "request_attempts": 32,
        "response_count": 32,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 907963,
        "uncached_input_tokens": 73227,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11828,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11868,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12001,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12049,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12398,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 18143,
            "output_tokens": 188,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 21983,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22311,
            "output_tokens": 169,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 25387,
            "output_tokens": 331,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26246,
            "output_tokens": 1147,
            "reasoning_output_tokens": 264
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27416,
            "output_tokens": 111,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28156,
            "output_tokens": 153,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28940,
            "output_tokens": 70,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29220,
            "output_tokens": 980,
            "reasoning_output_tokens": 181
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30222,
            "output_tokens": 119,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30972,
            "output_tokens": 54,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31088,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31185,
            "output_tokens": 119,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31705,
            "output_tokens": 118,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 32226,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32652,
            "output_tokens": 118,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33175,
            "output_tokens": 577,
            "reasoning_output_tokens": 455
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34138,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34606,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35030,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35499,
            "output_tokens": 197,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36097,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 36566,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36762,
            "output_tokens": 55,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 36879,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37071,
            "output_tokens": 731,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37824,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 47,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 15,
          "find": 4,
          "exec": 17,
          "devices": 1,
          "device": 2,
          "write": 3,
          "device_stop": 2,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 464 rosbridge control frames; 32 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 12
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/place-cans.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/workspace/.kinex/memos/place-cans.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_cartesian.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/workspace/.kinex/tools/dual_cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-33-20261116/workspace/.kinex/tools/dual_pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 82,
        "observed_images": 18
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 32,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 187,
        "published_frames": 187,
        "decode_ok": true,
        "duration_s": 4.675,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "982593474ddb42cd248a93b7423b5d92dcefeb471663e0704234e69a6ea0febf",
        "source_sha256": {
          "third_person": "e8458faa681f78d356f8d851d74582463891bb13cf69b67d5f83f660e3623cc6",
          "left_wrist": "2c77c9a604f94fd32406a5caf0f5943f382d41e08e59c4a1de65b6ad20a07980",
          "right_wrist": "9eff001d1409ed7dd7a293154ef29cd11cc88b08b1291e50676ba270a98ac604"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "de4d8bd2b000",
          "task": "robotwin/place-cans-plasticbox-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 187,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 187,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 187,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 187,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 187,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 187,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-34-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/34",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "34",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 13.84,
        "step_limit": 400,
        "steps": 346,
        "success": false,
        "valid": true
      },
      "steps": 346,
      "simulation_time_s": 13.84,
      "wall_time_s": 1153.059656,
      "timing": {
        "simulation_time_s": 13.84,
        "wall_time_s": 1153.059656,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the ceramic drinking cup with floral design on the gray plate for dining.",
      "native_instruction": "Place the ceramic drinking cup with floral design on the gray plate for dining.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9564519427348833,
        "cache_read_input_tokens": 1810048,
        "cache_reported_input_tokens": 1892461,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1892461,
        "cached_input_tokens": 1810048,
        "cost_usd": null,
        "fresh_input_tokens": 82413,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1892461,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1810048,
        "known_fresh_input_tokens": 82413,
        "known_input_tokens": 1892461,
        "known_output_tokens": 10012,
        "known_reasoning_output_tokens": 3195,
        "output_tokens": 10012,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3195,
        "reasoning_reported_output_tokens": 10012,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 52,
          "cache_write_input_tokens": 52,
          "cache_write_reported_input_tokens": 52,
          "cached_input_tokens": 52,
          "fresh_input_tokens": 52,
          "input_tokens": 52,
          "output_tokens": 52,
          "reasoning_output_tokens": 52,
          "reasoning_reported_output_tokens": 52
        },
        "request_attempts": 52,
        "response_count": 52,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1902473,
        "uncached_input_tokens": 82413,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11814,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11854,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11987,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12331,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12482,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 15716,
            "output_tokens": 174,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 19938,
            "output_tokens": 167,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 23644,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 23972,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24163,
            "output_tokens": 69,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 25271,
            "output_tokens": 284,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25950,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26358,
            "output_tokens": 279,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 28501,
            "output_tokens": 1000,
            "reasoning_output_tokens": 141
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29523,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 33028,
            "output_tokens": 340,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33571,
            "output_tokens": 111,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33710,
            "output_tokens": 105,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 35012,
            "output_tokens": 67,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35289,
            "output_tokens": 71,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35826,
            "output_tokens": 901,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 36748,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37239,
            "output_tokens": 115,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37416,
            "output_tokens": 391,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38008,
            "output_tokens": 195,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38621,
            "output_tokens": 464,
            "reasoning_output_tokens": 246
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39492,
            "output_tokens": 763,
            "reasoning_output_tokens": 680
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40663,
            "output_tokens": 51,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40765,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 40904,
            "output_tokens": 123,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41049,
            "output_tokens": 180,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41673,
            "output_tokens": 294,
            "reasoning_output_tokens": 214
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42383,
            "output_tokens": 95,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42540,
            "output_tokens": 327,
            "reasoning_output_tokens": 220
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 44060,
            "output_tokens": 278,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44605,
            "output_tokens": 164,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 45188,
            "output_tokens": 52,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45291,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45550,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45632,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 46128,
            "output_tokens": 419,
            "reasoning_output_tokens": 232
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 46710,
            "output_tokens": 209,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47608,
            "output_tokens": 60,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 48566,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48512,
            "input_tokens": 49222,
            "output_tokens": 140,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49280,
            "input_tokens": 49389,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49408,
            "input_tokens": 50216,
            "output_tokens": 192,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50304,
            "input_tokens": 50825,
            "output_tokens": 164,
            "reasoning_output_tokens": 127
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50816,
            "input_tokens": 51051,
            "output_tokens": 299,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51200,
            "input_tokens": 51377,
            "output_tokens": 155,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 51560,
            "output_tokens": 457,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51328,
            "input_tokens": 52042,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 59,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 18,
          "bash": 1,
          "exec": 21,
          "devices": 2,
          "device": 4,
          "write": 3,
          "edit": 4,
          "device_stop": 4,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 346 rosbridge control frames; 40 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 3,
          "direct_motion_exec_calls": 9
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-34-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 114,
        "observed_images": 24
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 40,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 140,
        "published_frames": 140,
        "decode_ok": true,
        "duration_s": 3.5,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f45926130e1ca92d053a3db71de1d3bae46e9c808878425c53807bdc91a3207a",
        "source_sha256": {
          "third_person": "83156c9232518c16ea6586a65fa5523c2bc7d737f69b0dc010e99dede659d70d",
          "left_wrist": "3ccd84360db3563011b39e519a53733a493be1ca32514feab9f9b10c06bbc087",
          "right_wrist": "ac202dd0930b6fa10c7aae4fa4a3356f7b03a1c27bf5d69ccf95b05968cef1b2"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "88100407c1f3",
          "task": "robotwin/place-container-plate-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 140,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 140,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13840000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 140,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 140,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13840000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 140,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 140,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13840000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-35-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/35",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "35",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 17.48,
        "step_limit": 500,
        "steps": 437,
        "success": true,
        "valid": true
      },
      "steps": 437,
      "simulation_time_s": 17.48,
      "wall_time_s": 945.732414,
      "timing": {
        "simulation_time_s": 17.48,
        "wall_time_s": 945.732414,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the small smooth cup onto the light gray coaster gently.",
      "native_instruction": "Place the small smooth cup onto the light gray coaster gently.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9660092598499979,
        "cache_read_input_tokens": 1196160,
        "cache_reported_input_tokens": 1238249,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1238249,
        "cached_input_tokens": 1196160,
        "cost_usd": null,
        "fresh_input_tokens": 42089,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1238249,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1196160,
        "known_fresh_input_tokens": 42089,
        "known_input_tokens": 1238249,
        "known_output_tokens": 6951,
        "known_reasoning_output_tokens": 1065,
        "output_tokens": 6951,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1065,
        "reasoning_reported_output_tokens": 6951,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 36,
          "cache_write_input_tokens": 36,
          "cache_write_reported_input_tokens": 36,
          "cached_input_tokens": 36,
          "fresh_input_tokens": 36,
          "input_tokens": 36,
          "output_tokens": 36,
          "reasoning_output_tokens": 36,
          "reasoning_reported_output_tokens": 36
        },
        "request_attempts": 36,
        "response_count": 36,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1245200,
        "uncached_input_tokens": 42089,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11988,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15584,
            "output_tokens": 247,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 19930,
            "output_tokens": 217,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 26973,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27301,
            "output_tokens": 169,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 30377,
            "output_tokens": 312,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31157,
            "output_tokens": 352,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31753,
            "output_tokens": 892,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32667,
            "output_tokens": 117,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33296,
            "output_tokens": 69,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33575,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 34604,
            "output_tokens": 733,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35358,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35918,
            "output_tokens": 308,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36671,
            "output_tokens": 376,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37074,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37454,
            "output_tokens": 78,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37744,
            "output_tokens": 68,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38507,
            "output_tokens": 56,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38585,
            "output_tokens": 291,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39159,
            "output_tokens": 309,
            "reasoning_output_tokens": 185
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39978,
            "output_tokens": 89,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40125,
            "output_tokens": 122,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40547,
            "output_tokens": 105,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40954,
            "output_tokens": 235,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41485,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41862,
            "output_tokens": 170,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42335,
            "output_tokens": 147,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42786,
            "output_tokens": 158,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43227,
            "output_tokens": 197,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43566,
            "output_tokens": 64,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 43692,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 43885,
            "output_tokens": 554,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44462,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 51,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 18,
          "find": 1,
          "exec": 18,
          "devices": 1,
          "device": 3,
          "write": 3,
          "edit": 1,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 437 rosbridge control frames; 40 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 9
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/cup_placement.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/workspace/.kinex/memos/cup_placement.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-35-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 90,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 40,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 176,
        "published_frames": 176,
        "decode_ok": true,
        "duration_s": 4.4,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "0a0d8c94a6daa20ed7fffa2c7c5ad726ff2feaa19f2eef78dc0fc7a218a164bb",
        "source_sha256": {
          "third_person": "11f2e7d4c9d247dfd1fccee784aa5a850114f07121b0161c886d99d4b7946a34",
          "left_wrist": "f3eeead1397c938910637687ee3606aa636bb72d4c70e5d06db77ed5621d0f03",
          "right_wrist": "fdf0e09613147289e261758e09381acdf2cda1c46797349a8f45ae41876b8f7e"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "bf0353d62e97",
          "task": "robotwin/place-empty-cup-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 176,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 176,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17480000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 176,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 176,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17480000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 176,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 176,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17480000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-36-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/36",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "36",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 14.16,
        "step_limit": 400,
        "steps": 354,
        "success": false,
        "valid": true
      },
      "steps": 354,
      "simulation_time_s": 14.16,
      "wall_time_s": 921.408177,
      "timing": {
        "simulation_time_s": 14.16,
        "wall_time_s": 921.408177,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the beige circular blade fan, place it on the Pink mat, face it toward the robot",
      "native_instruction": "Pick the beige circular blade fan, place it on the Pink mat, face it toward the robot",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9728440093146123,
        "cache_read_input_tokens": 1003904,
        "cache_reported_input_tokens": 1031927,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1031927,
        "cached_input_tokens": 1003904,
        "cost_usd": null,
        "fresh_input_tokens": 28023,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1031927,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1003904,
        "known_fresh_input_tokens": 28023,
        "known_input_tokens": 1031927,
        "known_output_tokens": 8632,
        "known_reasoning_output_tokens": 2595,
        "output_tokens": 8632,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2595,
        "reasoning_reported_output_tokens": 8632,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 36,
          "cache_write_input_tokens": 36,
          "cache_write_reported_input_tokens": 36,
          "cached_input_tokens": 36,
          "fresh_input_tokens": 36,
          "input_tokens": 36,
          "output_tokens": 36,
          "reasoning_output_tokens": 36,
          "reasoning_reported_output_tokens": 36
        },
        "request_attempts": 36,
        "response_count": 36,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1040559,
        "uncached_input_tokens": 28023,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11820,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11860,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11993,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12043,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12336,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16593,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 20409,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20737,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 22342,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22750,
            "output_tokens": 120,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23292,
            "output_tokens": 344,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24086,
            "output_tokens": 235,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25628,
            "output_tokens": 304,
            "reasoning_output_tokens": 264
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26398,
            "output_tokens": 371,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27192,
            "output_tokens": 1137,
            "reasoning_output_tokens": 272
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28351,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 29147,
            "output_tokens": 111,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29468,
            "output_tokens": 119,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30766,
            "output_tokens": 907,
            "reasoning_output_tokens": 104
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31694,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32117,
            "output_tokens": 110,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32677,
            "output_tokens": 116,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33140,
            "output_tokens": 109,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33594,
            "output_tokens": 497,
            "reasoning_output_tokens": 228
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 35290,
            "output_tokens": 764,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36572,
            "output_tokens": 170,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 36810,
            "output_tokens": 328,
            "reasoning_output_tokens": 239
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37204,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37646,
            "output_tokens": 250,
            "reasoning_output_tokens": 148
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38242,
            "output_tokens": 300,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38769,
            "output_tokens": 134,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39254,
            "output_tokens": 169,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39644,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40093,
            "output_tokens": 755,
            "reasoning_output_tokens": 270
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 40873,
            "output_tokens": 196,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41097,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 44,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "bash": 1,
          "devices": 1,
          "exec": 20,
          "write": 3,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 354 rosbridge control frames; 22 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 9
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-36-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 83,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 22,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 143,
        "published_frames": 143,
        "decode_ok": true,
        "duration_s": 3.575,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "fdb571984675dfc64ae839bfeb160e32b97ed8276ff04ed7db24366ac63b169f",
        "source_sha256": {
          "third_person": "413c0f151ea3ed962beb139955e9301f5cf05ff95966eae35631ecd5611d501c",
          "left_wrist": "5327d68ec2f40baf6eb98572e77df644a6c3dad7a0f6bb07093505b97bbcc80d",
          "right_wrist": "f7b11f44ed9ece7046bf488a1b04968a0d4fa158891d9c8f850dba4f4f7497a4"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "74a044290357",
          "task": "robotwin/place-fan-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 143,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 143,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14160000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 143,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 143,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14160000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 143,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 143,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14160000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-37-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/37",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "37",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 2
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 8.68,
        "step_limit": 400,
        "steps": 217,
        "success": true,
        "valid": true
      },
      "steps": 217,
      "simulation_time_s": 8.68,
      "wall_time_s": 443.230361,
      "timing": {
        "simulation_time_s": 8.68,
        "wall_time_s": 443.230361,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the dark gray computer mouse with logo with the left arm and drop it onto the Cyan mat.",
      "native_instruction": "Pick up the dark gray computer mouse with logo with the left arm and drop it onto the Cyan mat.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9466529274304385,
        "cache_read_input_tokens": 710144,
        "cache_reported_input_tokens": 750163,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 750163,
        "cached_input_tokens": 710144,
        "cost_usd": null,
        "fresh_input_tokens": 40019,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 750163,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 710144,
        "known_fresh_input_tokens": 40019,
        "known_input_tokens": 750163,
        "known_output_tokens": 4476,
        "known_reasoning_output_tokens": 815,
        "output_tokens": 4476,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 815,
        "reasoning_reported_output_tokens": 4476,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 36,
          "cache_write_input_tokens": 36,
          "cache_write_reported_input_tokens": 36,
          "cached_input_tokens": 36,
          "fresh_input_tokens": 36,
          "input_tokens": 36,
          "output_tokens": 36,
          "reasoning_output_tokens": 36,
          "reasoning_reported_output_tokens": 36
        },
        "request_attempts": 36,
        "response_count": 36,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 754639,
        "uncached_input_tokens": 40019,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11824,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11859,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11992,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12040,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12333,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 12484,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 12947,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12800,
            "input_tokens": 13761,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13696,
            "input_tokens": 14587,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 15379,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15232,
            "input_tokens": 16604,
            "output_tokens": 90,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 17902,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18472,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 18636,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 18826,
            "output_tokens": 66,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 19102,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 20181,
            "output_tokens": 312,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20888,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 21296,
            "output_tokens": 309,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 22668,
            "output_tokens": 83,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 23217,
            "output_tokens": 927,
            "reasoning_output_tokens": 142
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 24165,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24865,
            "output_tokens": 64,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 24956,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25282,
            "output_tokens": 113,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25705,
            "output_tokens": 127,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26146,
            "output_tokens": 106,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26274,
            "output_tokens": 215,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26784,
            "output_tokens": 144,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27246,
            "output_tokens": 296,
            "reasoning_output_tokens": 205
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27860,
            "output_tokens": 121,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 28289,
            "output_tokens": 149,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28580,
            "output_tokens": 57,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28675,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28867,
            "output_tokens": 579,
            "reasoning_output_tokens": 108
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29471,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 35,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 12,
          "bash": 1,
          "exec": 12,
          "devices": 1,
          "device": 2,
          "write": 2,
          "edit": 1,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 217 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 9
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/move_left.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-37-20261116/workspace/.kinex/tools/move_left.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 74,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 88,
        "published_frames": 88,
        "decode_ok": true,
        "duration_s": 2.2,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "d96a7119717e112e83ebccf4c06d3ae79eedf827be02bd97ab6cdaf9e9d4946d",
        "source_sha256": {
          "third_person": "94be867e639a8f45f14b96f145eee790b17633f329434bdf0b36fae0037da94e",
          "left_wrist": "52038722b633d715064b017f921e3e7b948acf1405d2c341ddb85f24753c9ab0",
          "right_wrist": "79c40aba8019beae98c91af2287ca64fbb51bcf7ce011f36b2d9fb8fda02e6c1"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "fe618643602d",
          "task": "robotwin/place-mouse-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 88,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 88,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8680000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 88,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 88,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8680000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 88,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 88,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8680000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": false
    },
    {
      "id": "onboard-kinex-tcp-posestate-38-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/38",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "38",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 24.400000000000002,
        "step_limit": 700,
        "steps": 610,
        "success": false,
        "valid": true
      },
      "steps": 610,
      "simulation_time_s": 24.400000000000002,
      "wall_time_s": 1165.071046,
      "timing": {
        "simulation_time_s": 24.400000000000002,
        "wall_time_s": 1165.071046,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the toycar with rounded edges and place into the basket with rounded edges.",
      "native_instruction": "Grab the toycar with rounded edges and place into the basket with rounded edges.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9668759993722293,
        "cache_read_input_tokens": 1275264,
        "cache_reported_input_tokens": 1318953,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1318953,
        "cached_input_tokens": 1275264,
        "cost_usd": null,
        "fresh_input_tokens": 43689,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1318953,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1275264,
        "known_fresh_input_tokens": 43689,
        "known_input_tokens": 1318953,
        "known_output_tokens": 7497,
        "known_reasoning_output_tokens": 1779,
        "output_tokens": 7497,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1779,
        "reasoning_reported_output_tokens": 7497,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 41,
          "cache_write_input_tokens": 41,
          "cache_write_reported_input_tokens": 41,
          "cached_input_tokens": 41,
          "fresh_input_tokens": 41,
          "input_tokens": 41,
          "output_tokens": 41,
          "reasoning_output_tokens": 41,
          "reasoning_reported_output_tokens": 41
        },
        "request_attempts": 41,
        "response_count": 41,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1326450,
        "uncached_input_tokens": 43689,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11820,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11860,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11993,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14697,
            "output_tokens": 201,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14720,
            "input_tokens": 17451,
            "output_tokens": 245,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 22885,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23213,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 24613,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24805,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25884,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26292,
            "output_tokens": 281,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 27089,
            "output_tokens": 1157,
            "reasoning_output_tokens": 304
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28271,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 29096,
            "output_tokens": 131,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29967,
            "output_tokens": 154,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 31782,
            "output_tokens": 63,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31907,
            "output_tokens": 795,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32725,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33256,
            "output_tokens": 103,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33798,
            "output_tokens": 60,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33909,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34048,
            "output_tokens": 76,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34146,
            "output_tokens": 197,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34529,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35303,
            "output_tokens": 59,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 36283,
            "output_tokens": 161,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36473,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37299,
            "output_tokens": 198,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37526,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38515,
            "output_tokens": 139,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39097,
            "output_tokens": 301,
            "reasoning_output_tokens": 186
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 40299,
            "output_tokens": 152,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40893,
            "output_tokens": 77,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 40992,
            "output_tokens": 410,
            "reasoning_output_tokens": 153
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41797,
            "output_tokens": 458,
            "reasoning_output_tokens": 362
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42704,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43240,
            "output_tokens": 173,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43856,
            "output_tokens": 216,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44293,
            "output_tokens": 105,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44838,
            "output_tokens": 645,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45509,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 54,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 18,
          "find": 1,
          "devices": 1,
          "exec": 19,
          "device": 2,
          "write": 3,
          "device_stop": 3,
          "relative_pose": 1,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 610 rosbridge control frames; 60 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 14
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin_basket.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/workspace/.kinex/memos/robotwin_basket.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-38-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 98,
        "observed_images": 25
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 60,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 245,
        "published_frames": 245,
        "decode_ok": true,
        "duration_s": 6.125,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "323c3af5b0f291e6a9433349aa80058aa9b12b0f6acd6ba45163d7c7cd8efca8",
        "source_sha256": {
          "third_person": "e4c3af9222f6d796b9c6bceb24838e7c920e0a67b6ed1310ca4387ddbd6560fb",
          "left_wrist": "f68b4938b057a2f58668a249101b774f341f2fe689d97f7682629a635af8c8af",
          "right_wrist": "5a4f33dcf8126c1153df41374cb9836a88b579a0371595724a6f7e00e64ef5ef"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "aeba4c1b8eb1",
          "task": "robotwin/place-object-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 246,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 245,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24400000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 246,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 245,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24400000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 246,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 245,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24400000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-39-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/39",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "39",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 8.44,
        "step_limit": 400,
        "steps": 211,
        "success": true,
        "valid": true
      },
      "steps": 211,
      "simulation_time_s": 8.44,
      "wall_time_s": 354.189808,
      "timing": {
        "simulation_time_s": 8.44,
        "wall_time_s": 354.189808,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use one arm to pick the palm-sized black stapler for the black screen electronicscale",
      "native_instruction": "Use one arm to pick the palm-sized black stapler for the black screen electronicscale",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9682640339292675,
        "cache_read_input_tokens": 665728,
        "cache_reported_input_tokens": 687548,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 687548,
        "cached_input_tokens": 665728,
        "cost_usd": null,
        "fresh_input_tokens": 21820,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 687548,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 665728,
        "known_fresh_input_tokens": 21820,
        "known_input_tokens": 687548,
        "known_output_tokens": 3849,
        "known_reasoning_output_tokens": 474,
        "output_tokens": 3849,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 474,
        "reasoning_reported_output_tokens": 3849,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 29,
          "cache_write_input_tokens": 29,
          "cache_write_reported_input_tokens": 29,
          "cached_input_tokens": 29,
          "fresh_input_tokens": 29,
          "input_tokens": 29,
          "output_tokens": 29,
          "reasoning_output_tokens": 29,
          "reasoning_reported_output_tokens": 29
        },
        "request_attempts": 29,
        "response_count": 29,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 691397,
        "uncached_input_tokens": 21820,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11951,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12003,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12352,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18097,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18667,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 18897,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19088,
            "output_tokens": 174,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 21389,
            "output_tokens": 311,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 22095,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22503,
            "output_tokens": 288,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23896,
            "output_tokens": 141,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24503,
            "output_tokens": 87,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25769,
            "output_tokens": 797,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26587,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26726,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27115,
            "output_tokens": 111,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27531,
            "output_tokens": 71,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27624,
            "output_tokens": 60,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27702,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27784,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28053,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28443,
            "output_tokens": 163,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28911,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29301,
            "output_tokens": 64,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29383,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29432,
            "output_tokens": 60,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29720,
            "output_tokens": 463,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30208,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 40,
        "model_tool_calls_by_name": {
          "read": 15,
          "ls": 1,
          "find": 4,
          "devices": 3,
          "device": 5,
          "exec": 9,
          "write": 2,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 211 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 5
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-39-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 72,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 86,
        "published_frames": 86,
        "decode_ok": true,
        "duration_s": 2.15,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "5ba2beea380033bb26e6489db083ee8f03d0fd258e5921e430b6ccdd5f08f12a",
        "source_sha256": {
          "third_person": "43e0ff588b72bf8d724eca5d469fc3f1a747deb379f92681d5c7659f4c3de8d5",
          "left_wrist": "19dd8038af93417ec17c7b28e86eff31fbc826d3869f4f4dad940fc3fbc919ec",
          "right_wrist": "c090952dc6dfbc8c2baa5c305096b570502e827774447f240a915e08c9abb46b"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "fee92a65d766",
          "task": "robotwin/place-object-scale-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 86,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 86,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8440000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 86,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 86,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8440000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 86,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 86,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8440000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": false
    },
    {
      "id": "onboard-kinex-tcp-posestate-40-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/40",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "40",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 11.68,
        "step_limit": 400,
        "steps": 292,
        "success": true,
        "valid": true
      },
      "steps": 292,
      "simulation_time_s": 11.68,
      "wall_time_s": 784.221101,
      "timing": {
        "simulation_time_s": 11.68,
        "wall_time_s": 784.221101,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set the rectangular remotecontrol on the displaystand with hollow center.",
      "native_instruction": "Set the rectangular remotecontrol on the displaystand with hollow center.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9316201491660353,
        "cache_read_input_tokens": 745216,
        "cache_reported_input_tokens": 799914,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 799914,
        "cached_input_tokens": 745216,
        "cost_usd": null,
        "fresh_input_tokens": 54698,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 799914,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 745216,
        "known_fresh_input_tokens": 54698,
        "known_input_tokens": 799914,
        "known_output_tokens": 5462,
        "known_reasoning_output_tokens": 1064,
        "output_tokens": 5462,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1064,
        "reasoning_reported_output_tokens": 5462,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 31,
          "cache_write_input_tokens": 31,
          "cache_write_reported_input_tokens": 31,
          "cached_input_tokens": 31,
          "fresh_input_tokens": 31,
          "input_tokens": 31,
          "output_tokens": 31,
          "reasoning_output_tokens": 31,
          "reasoning_reported_output_tokens": 31
        },
        "request_attempts": 31,
        "response_count": 31,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 805376,
        "uncached_input_tokens": 54698,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11949,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11997,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12290,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12441,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 18186,
            "output_tokens": 189,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 21779,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22107,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22298,
            "output_tokens": 117,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23821,
            "output_tokens": 289,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24627,
            "output_tokens": 124,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25217,
            "output_tokens": 884,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26123,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26288,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 27010,
            "output_tokens": 210,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27839,
            "output_tokens": 69,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28118,
            "output_tokens": 106,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 29403,
            "output_tokens": 687,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30111,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30502,
            "output_tokens": 76,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30629,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30889,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30976,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31378,
            "output_tokens": 545,
            "reasoning_output_tokens": 422
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32226,
            "output_tokens": 265,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32648,
            "output_tokens": 173,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33127,
            "output_tokens": 44,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33189,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33238,
            "output_tokens": 61,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33529,
            "output_tokens": 609,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34161,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 40,
        "model_tool_calls_by_name": {
          "read": 17,
          "ls": 1,
          "bash": 1,
          "exec": 10,
          "devices": 3,
          "device": 5,
          "write": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 292 rosbridge control frames; 36 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 4
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/place_object_stand.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/workspace/.kinex/memos/place_object_stand.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-40-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 74,
        "observed_images": 12
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 36,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 118,
        "published_frames": 118,
        "decode_ok": true,
        "duration_s": 2.95,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "89c0ba0d352a16bb270d46b2b4a59e49934d8d3f6251bf53447423024133d67a",
        "source_sha256": {
          "third_person": "c804769f3b4a38566893f41f7444e189ecd706954f69bd466ea7a810d4be0922",
          "left_wrist": "9e077dc648355bb3cf8bf263acf5ac8c5def21a637f9d56744229a06cace5a19",
          "right_wrist": "08e062538d13a4fc23df1773a0bd55f0c8d3ef675a6576128d2e07330ff14ef4"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "fafbc071db96",
          "task": "robotwin/place-object-stand-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 118,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 118,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11680000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 118,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 118,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11680000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 118,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 118,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11680000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-41-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/41",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "41",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.76,
        "step_limit": 400,
        "steps": 394,
        "success": false,
        "valid": true
      },
      "steps": 394,
      "simulation_time_s": 15.76,
      "wall_time_s": 1041.124084,
      "timing": {
        "simulation_time_s": 15.76,
        "wall_time_s": 1041.124084,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set the phone with shiny black back on the L-shaped smooth brown phonestand.",
      "native_instruction": "Set the phone with shiny black back on the L-shaped smooth brown phonestand.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9689058979524628,
        "cache_read_input_tokens": 1226880,
        "cache_reported_input_tokens": 1266253,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1266253,
        "cached_input_tokens": 1226880,
        "cost_usd": null,
        "fresh_input_tokens": 39373,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1266253,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1226880,
        "known_fresh_input_tokens": 39373,
        "known_input_tokens": 1266253,
        "known_output_tokens": 8460,
        "known_reasoning_output_tokens": 2603,
        "output_tokens": 8460,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2603,
        "reasoning_reported_output_tokens": 8460,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 40,
          "cache_write_input_tokens": 40,
          "cache_write_reported_input_tokens": 40,
          "cached_input_tokens": 40,
          "fresh_input_tokens": 40,
          "input_tokens": 40,
          "output_tokens": 40,
          "reasoning_output_tokens": 40,
          "reasoning_reported_output_tokens": 40
        },
        "request_attempts": 40,
        "response_count": 40,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1274713,
        "uncached_input_tokens": 39373,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11988,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14735,
            "output_tokens": 169,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14720,
            "input_tokens": 16679,
            "output_tokens": 215,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16768,
            "input_tokens": 21335,
            "output_tokens": 93,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 22926,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23254,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23662,
            "output_tokens": 134,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24902,
            "output_tokens": 257,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25609,
            "output_tokens": 91,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26166,
            "output_tokens": 1058,
            "reasoning_output_tokens": 195
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27248,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 31195,
            "output_tokens": 97,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31502,
            "output_tokens": 855,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32380,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32734,
            "output_tokens": 140,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33169,
            "output_tokens": 67,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33258,
            "output_tokens": 178,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33624,
            "output_tokens": 91,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33994,
            "output_tokens": 520,
            "reasoning_output_tokens": 227
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34927,
            "output_tokens": 579,
            "reasoning_output_tokens": 505
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35574,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35925,
            "output_tokens": 424,
            "reasoning_output_tokens": 204
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36573,
            "output_tokens": 194,
            "reasoning_output_tokens": 116
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37043,
            "output_tokens": 146,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37482,
            "output_tokens": 205,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37876,
            "output_tokens": 63,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 37961,
            "output_tokens": 122,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38362,
            "output_tokens": 396,
            "reasoning_output_tokens": 156
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39144,
            "output_tokens": 390,
            "reasoning_output_tokens": 314
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39813,
            "output_tokens": 165,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40162,
            "output_tokens": 152,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40591,
            "output_tokens": 197,
            "reasoning_output_tokens": 122
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41066,
            "output_tokens": 186,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41446,
            "output_tokens": 69,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41537,
            "output_tokens": 169,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41984,
            "output_tokens": 95,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42141,
            "output_tokens": 450,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42616,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 49,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 13,
          "find": 1,
          "exec": 23,
          "devices": 1,
          "write": 3,
          "device_stop": 4
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 394 rosbridge control frames; 20 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 13
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/place_phone_stand.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/workspace/.kinex/memos/place_phone_stand.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-41-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 92,
        "observed_images": 17
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 20,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 159,
        "published_frames": 159,
        "decode_ok": true,
        "duration_s": 3.975,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "b770e8d2c486e06ea5eed7c87ca38998fddd97f07826f9a50d87b9afd93ed1f4",
        "source_sha256": {
          "third_person": "9c9f4aa2a6dc79599e3299d2293b3720af93adf7eb7c1b8c1f8845002c2ec228",
          "left_wrist": "3bfb275bac5c0b7fc391342e073e882ebeb563c5ead0e3df4c8f042083cbb2f7",
          "right_wrist": "5218a4883c8e9d64422b8d529e72fc8d6c137bc164c7702c936f62a355591f91"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "18a291c767f9",
          "task": "robotwin/place-phone-stand-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 159,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 159,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15760000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 159,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 159,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15760000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 159,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 159,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15760000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-42-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/42",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "42",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 13.48,
        "step_limit": 500,
        "steps": 337,
        "success": true,
        "valid": true
      },
      "steps": 337,
      "simulation_time_s": 13.48,
      "wall_time_s": 902.275125,
      "timing": {
        "simulation_time_s": 13.48,
        "wall_time_s": 902.275125,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the casual brown shoe off the table and drop it on the mat",
      "native_instruction": "Grab the casual brown shoe off the table and drop it on the mat",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9197323072901096,
        "cache_read_input_tokens": 710656,
        "cache_reported_input_tokens": 772677,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 772677,
        "cached_input_tokens": 710656,
        "cost_usd": null,
        "fresh_input_tokens": 62021,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 772677,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 710656,
        "known_fresh_input_tokens": 62021,
        "known_input_tokens": 772677,
        "known_output_tokens": 4729,
        "known_reasoning_output_tokens": 578,
        "output_tokens": 4729,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 578,
        "reasoning_reported_output_tokens": 4729,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 26,
          "cache_write_input_tokens": 26,
          "cache_write_reported_input_tokens": 26,
          "cached_input_tokens": 26,
          "fresh_input_tokens": 26,
          "input_tokens": 26,
          "output_tokens": 26,
          "reasoning_output_tokens": 26,
          "reasoning_reported_output_tokens": 26
        },
        "request_attempts": 26,
        "response_count": 26,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 777406,
        "uncached_input_tokens": 62021,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11817,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11857,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11990,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15482,
            "output_tokens": 206,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 19751,
            "output_tokens": 240,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15488,
            "input_tokens": 23670,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 25640,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 25968,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26161,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 27240,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27648,
            "output_tokens": 393,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 29812,
            "output_tokens": 877,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30713,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30874,
            "output_tokens": 142,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 32594,
            "output_tokens": 183,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33882,
            "output_tokens": 152,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 35139,
            "output_tokens": 181,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 36587,
            "output_tokens": 116,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37900,
            "output_tokens": 741,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38664,
            "output_tokens": 132,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39259,
            "output_tokens": 175,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 39576,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39667,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39934,
            "output_tokens": 241,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40205,
            "output_tokens": 418,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 40647,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 38,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 16,
          "find": 1,
          "exec": 9,
          "devices": 2,
          "device": 1,
          "write": 3,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 337 rosbridge control frames; 74 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 4,
          "direct_motion_exec_calls": 2
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/place-shoe.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/workspace/.kinex/memos/place-shoe.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-42-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 67,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 74,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 136,
        "published_frames": 136,
        "decode_ok": true,
        "duration_s": 3.4,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "1252792158a31aa14dced65e10a3768a8d618c0b8f532eab6df8dfd89112757b",
        "source_sha256": {
          "third_person": "8b058a1da3f59fd9e3f901f37ad7bd5f8bf1926913baf09c1438d602b2a55264",
          "left_wrist": "9f59e8e5f9c25821ee756309f1c38fb2f1842f7a86ca1a0fde6417b5ed21c298",
          "right_wrist": "537f46aa5341eb2572371c2c4af6df0e314e99edc9ef98285bcae7a969ed825a"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "c6ab1fd5dd01",
          "task": "robotwin/place-shoe-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 136,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 136,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13480000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 136,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 136,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13480000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 136,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 136,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13480000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-43-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/43",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "43",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 4.5600000000000005,
        "step_limit": 400,
        "steps": 114,
        "success": true,
        "valid": true
      },
      "steps": 114,
      "simulation_time_s": 4.5600000000000005,
      "wall_time_s": 307.860484,
      "timing": {
        "simulation_time_s": 4.5600000000000005,
        "wall_time_s": 307.860484,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to press the shiny red plastic stapler.",
      "native_instruction": "Use the left arm to press the shiny red plastic stapler.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9465661534047798,
        "cache_read_input_tokens": 501504,
        "cache_reported_input_tokens": 529814,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 529814,
        "cached_input_tokens": 501504,
        "cost_usd": null,
        "fresh_input_tokens": 28310,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 529814,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 501504,
        "known_fresh_input_tokens": 28310,
        "known_input_tokens": 529814,
        "known_output_tokens": 3925,
        "known_reasoning_output_tokens": 448,
        "output_tokens": 3925,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 448,
        "reasoning_reported_output_tokens": 3925,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 25,
          "cache_write_input_tokens": 25,
          "cache_write_reported_input_tokens": 25,
          "cached_input_tokens": 25,
          "fresh_input_tokens": 25,
          "input_tokens": 25,
          "output_tokens": 25,
          "reasoning_output_tokens": 25,
          "reasoning_reported_output_tokens": 25
        },
        "request_attempts": 25,
        "response_count": 25,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 533739,
        "uncached_input_tokens": 28310,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11817,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11857,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11990,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12040,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13500,
            "output_tokens": 174,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13568,
            "input_tokens": 16548,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 17810,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 17974,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18990,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19242,
            "output_tokens": 439,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20540,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20948,
            "output_tokens": 356,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 22129,
            "output_tokens": 996,
            "reasoning_output_tokens": 214
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23148,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 24227,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24909,
            "output_tokens": 102,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25040,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25411,
            "output_tokens": 83,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26189,
            "output_tokens": 86,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26700,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26916,
            "output_tokens": 85,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27412,
            "output_tokens": 114,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27737,
            "output_tokens": 349,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28115,
            "output_tokens": 484,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28625,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 31,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 13,
          "find": 1,
          "exec": 7,
          "devices": 2,
          "device": 2,
          "write": 2,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 114 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 4
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/press-stapler.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/workspace/.kinex/memos/press-stapler.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_left.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-43-20261116/workspace/.kinex/tools/cartesian_left.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 59,
        "observed_images": 3
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 47,
        "published_frames": 47,
        "decode_ok": true,
        "duration_s": 1.175,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "6f41e0eaf89f568fa32ec555d505effea754fc29a12adf1afa980e6ddbe2b318",
        "source_sha256": {
          "third_person": "c5b2e76a82f676b26ddb0465db09f1615f0c686ecac4e4b23487b4ef1e50c8be",
          "left_wrist": "d43a0199e010a7dad02c1f558fc7de675aea1274146c286c15bf0e903200f536",
          "right_wrist": "5771b293929675e9b0c656bd2398dfa787616d09493ed4af46dfa076693e8aa9"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "4700732c2a75",
          "task": "robotwin/press-stapler-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 47,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 47,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 4560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 47,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 47,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 4560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 47,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 47,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 4560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": false
    },
    {
      "id": "onboard-kinex-tcp-posestate-44-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/44",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "44",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.44,
        "step_limit": 400,
        "steps": 386,
        "success": false,
        "valid": true
      },
      "steps": 386,
      "simulation_time_s": 15.44,
      "wall_time_s": 1075.430641,
      "timing": {
        "simulation_time_s": 15.44,
        "wall_time_s": 1075.430641,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Catch the tabletop paymentsign, pick it up and rotate it, then set it back down.",
      "native_instruction": "Catch the tabletop paymentsign, pick it up and rotate it, then set it back down.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9772624887995603,
        "cache_read_input_tokens": 1579264,
        "cache_reported_input_tokens": 1616008,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1616008,
        "cached_input_tokens": 1579264,
        "cost_usd": null,
        "fresh_input_tokens": 36744,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1616008,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1579264,
        "known_fresh_input_tokens": 36744,
        "known_input_tokens": 1616008,
        "known_output_tokens": 9897,
        "known_reasoning_output_tokens": 1970,
        "output_tokens": 9897,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1970,
        "reasoning_reported_output_tokens": 9897,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 46,
          "cache_write_input_tokens": 46,
          "cache_write_reported_input_tokens": 46,
          "cached_input_tokens": 46,
          "fresh_input_tokens": 46,
          "input_tokens": 46,
          "output_tokens": 46,
          "reasoning_output_tokens": 46,
          "reasoning_reported_output_tokens": 46
        },
        "request_attempts": 46,
        "response_count": 46,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1625905,
        "uncached_input_tokens": 36744,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11822,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11862,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11995,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15535,
            "output_tokens": 157,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15488,
            "input_tokens": 17467,
            "output_tokens": 228,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 23133,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23703,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24031,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 25430,
            "output_tokens": 141,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25966,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26374,
            "output_tokens": 122,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26920,
            "output_tokens": 340,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27656,
            "output_tokens": 306,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28637,
            "output_tokens": 399,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29284,
            "output_tokens": 437,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30080,
            "output_tokens": 129,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30675,
            "output_tokens": 986,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31685,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31908,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 36417,
            "output_tokens": 72,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36551,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36803,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37832,
            "output_tokens": 881,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 38736,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39175,
            "output_tokens": 113,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39639,
            "output_tokens": 142,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40113,
            "output_tokens": 170,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40555,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40988,
            "output_tokens": 234,
            "reasoning_output_tokens": 148
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41482,
            "output_tokens": 106,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41938,
            "output_tokens": 59,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42019,
            "output_tokens": 154,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42580,
            "output_tokens": 295,
            "reasoning_output_tokens": 209
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43130,
            "output_tokens": 717,
            "reasoning_output_tokens": 435
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 44015,
            "output_tokens": 526,
            "reasoning_output_tokens": 145
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 45039,
            "output_tokens": 203,
            "reasoning_output_tokens": 117
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45595,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 46052,
            "output_tokens": 172,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46491,
            "output_tokens": 163,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 47003,
            "output_tokens": 243,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 47488,
            "output_tokens": 204,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 47917,
            "output_tokens": 70,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47872,
            "input_tokens": 48049,
            "output_tokens": 294,
            "reasoning_output_tokens": 133
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 48373,
            "output_tokens": 156,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 48558,
            "output_tokens": 724,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49152,
            "input_tokens": 49307,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 56,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 16,
          "find": 1,
          "devices": 1,
          "exec": 26,
          "write": 3,
          "device_stop": 3,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 386 rosbridge control frames; 20 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 11
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/rotate-qrcode.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/workspace/.kinex/memos/rotate-qrcode.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-44-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 105,
        "observed_images": 22
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 20,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 156,
        "published_frames": 156,
        "decode_ok": true,
        "duration_s": 3.9,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "07b1d1b046650468411b6de764d659135d4dd5a73b97db9bd8e4db1dd6045a18",
        "source_sha256": {
          "third_person": "89c26bc32625c6ea4f45bde4cce02e910f2d912dc9299d4876763f7f969d5626",
          "left_wrist": "dc05ce06f4b70fd8b744d933959552540f4748d62c5a4a051098ff38978da75a",
          "right_wrist": "6af2f62b07fc085a9149ca9ebf14137ad9a8693e2e963819081cb74ca6b41eb7"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "82f7ebff3eb6",
          "task": "robotwin/rotate-qrcode-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 156,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 156,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15440000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 156,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 156,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15440000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 156,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 156,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15440000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-45-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/45",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "45",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.24,
        "step_limit": 700,
        "steps": 231,
        "success": true,
        "valid": true
      },
      "steps": 231,
      "simulation_time_s": 9.24,
      "wall_time_s": 662.91672,
      "timing": {
        "simulation_time_s": 9.24,
        "wall_time_s": 662.91672,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the bottle with white cap with the left arm and shake.",
      "native_instruction": "Pick up the bottle with white cap with the left arm and shake.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.959454586848307,
        "cache_read_input_tokens": 779648,
        "cache_reported_input_tokens": 812595,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 812595,
        "cached_input_tokens": 779648,
        "cost_usd": null,
        "fresh_input_tokens": 32947,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 812595,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 779648,
        "known_fresh_input_tokens": 32947,
        "known_input_tokens": 812595,
        "known_output_tokens": 4922,
        "known_reasoning_output_tokens": 657,
        "output_tokens": 4922,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 657,
        "reasoning_reported_output_tokens": 4922,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 29,
          "cache_write_input_tokens": 29,
          "cache_write_reported_input_tokens": 29,
          "cached_input_tokens": 29,
          "fresh_input_tokens": 29,
          "input_tokens": 29,
          "output_tokens": 29,
          "reasoning_output_tokens": 29,
          "reasoning_reported_output_tokens": 29
        },
        "request_attempts": 29,
        "response_count": 29,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 817517,
        "uncached_input_tokens": 32947,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11988,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15066,
            "output_tokens": 201,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 18597,
            "output_tokens": 202,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 23199,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23527,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24926,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25382,
            "output_tokens": 104,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26525,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26933,
            "output_tokens": 273,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27737,
            "output_tokens": 89,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28292,
            "output_tokens": 919,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29233,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29397,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30458,
            "output_tokens": 74,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30742,
            "output_tokens": 74,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31805,
            "output_tokens": 350,
            "reasoning_output_tokens": 148
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32399,
            "output_tokens": 768,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33188,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33751,
            "output_tokens": 119,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34346,
            "output_tokens": 154,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34759,
            "output_tokens": 133,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35578,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35637,
            "output_tokens": 141,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35920,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36015,
            "output_tokens": 73,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36488,
            "output_tokens": 525,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37037,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 39,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 17,
          "find": 1,
          "devices": 3,
          "exec": 10,
          "write": 3,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 231 rosbridge control frames; 28 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 5
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/bottle_manipulation.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/workspace/.kinex/memos/bottle_manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-45-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 71,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 28,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 94,
        "published_frames": 94,
        "decode_ok": true,
        "duration_s": 2.35,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f4a517d85ab4b8af4a6d93e7ac5d7f6f1a6e3ab1f3dc69de2c9a15733782622a",
        "source_sha256": {
          "third_person": "458771d30b1df04dcf54841a31dda62fad1cdb1aad332e2ee11984e756cee2fc",
          "left_wrist": "6ce34a857e3e66ca880b5f205bcde730cb9a4501799d01863f2d199fd11d5752",
          "right_wrist": "ed1bc53e57b948fb00cd37cfead2dbed13a3ebab3c3156496bea57a974b129a0"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "0d545882d2c2",
          "task": "robotwin/shake-bottle-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 94,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 94,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9240000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 94,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 94,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9240000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 94,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 94,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9240000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-46-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/46",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "46",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 8.56,
        "step_limit": 700,
        "steps": 214,
        "success": true,
        "valid": true
      },
      "steps": 214,
      "simulation_time_s": 8.56,
      "wall_time_s": 641.298832,
      "timing": {
        "simulation_time_s": 8.56,
        "wall_time_s": 641.298832,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the bottle with white cap and shake it horizontally.",
      "native_instruction": "Grab the bottle with white cap and shake it horizontally.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9481784107009785,
        "cache_read_input_tokens": 683264,
        "cache_reported_input_tokens": 720607,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 720607,
        "cached_input_tokens": 683264,
        "cost_usd": null,
        "fresh_input_tokens": 37343,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 720607,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 683264,
        "known_fresh_input_tokens": 37343,
        "known_input_tokens": 720607,
        "known_output_tokens": 4599,
        "known_reasoning_output_tokens": 618,
        "output_tokens": 4599,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 618,
        "reasoning_reported_output_tokens": 4599,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 25,
          "cache_write_input_tokens": 25,
          "cache_write_reported_input_tokens": 25,
          "cached_input_tokens": 25,
          "fresh_input_tokens": 25,
          "input_tokens": 25,
          "output_tokens": 25,
          "reasoning_output_tokens": 25,
          "reasoning_reported_output_tokens": 25
        },
        "request_attempts": 25,
        "response_count": 25,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 725206,
        "uncached_input_tokens": 37343,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11820,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11860,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11993,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15567,
            "output_tokens": 288,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 21516,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 22086,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 24055,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24383,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24575,
            "output_tokens": 195,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 27764,
            "output_tokens": 318,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 29775,
            "output_tokens": 68,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30309,
            "output_tokens": 1005,
            "reasoning_output_tokens": 184
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31336,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 32048,
            "output_tokens": 78,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32336,
            "output_tokens": 175,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 34677,
            "output_tokens": 861,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35559,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35953,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36339,
            "output_tokens": 179,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36742,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37512,
            "output_tokens": 123,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37775,
            "output_tokens": 52,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 37889,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38081,
            "output_tokens": 552,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38657,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 37,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 18,
          "find": 1,
          "exec": 8,
          "devices": 1,
          "device": 2,
          "write": 3,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 214 rosbridge control frames; 30 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 5
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/aloha-bottle.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/workspace/.kinex/memos/aloha-bottle.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-46-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 65,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 30,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 87,
        "published_frames": 87,
        "decode_ok": true,
        "duration_s": 2.175,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "bd131dcf8fa0e9fde61da69c1b01b800a42b34581f767190093ecf04714303b5",
        "source_sha256": {
          "third_person": "565ffebe32ef442dd832f8fd2e42a39ffb467064a6dec9ea1523311ace9b2e3a",
          "left_wrist": "24a61ea17df0ac6c425f4b86c5e9a683722fdcceb7f0b06a30a0f5cf66c127a4",
          "right_wrist": "2237dc8e5f4e89e78480fdd82c69199cd980e430beb46cfa4a8a1b8e52f8112f"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "45aeb7ee3824",
          "task": "robotwin/shake-bottle-horizontally-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 87,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 87,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 87,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 87,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 87,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 87,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-47-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/47",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "47",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 23.0,
        "step_limit": 800,
        "steps": 575,
        "success": true,
        "valid": true
      },
      "steps": 575,
      "simulation_time_s": 23.0,
      "wall_time_s": 1103.7306,
      "timing": {
        "simulation_time_s": 23.0,
        "wall_time_s": 1103.7306,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Bring red block and green block to the center, then position green block above red block.",
      "native_instruction": "Bring red block and green block to the center, then position green block above red block.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9736427556775169,
        "cache_read_input_tokens": 1446912,
        "cache_reported_input_tokens": 1486081,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1486081,
        "cached_input_tokens": 1446912,
        "cost_usd": null,
        "fresh_input_tokens": 39169,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1486081,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1446912,
        "known_fresh_input_tokens": 39169,
        "known_input_tokens": 1486081,
        "known_output_tokens": 7583,
        "known_reasoning_output_tokens": 1590,
        "output_tokens": 7583,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1590,
        "reasoning_reported_output_tokens": 7583,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 42,
          "cache_write_input_tokens": 42,
          "cache_write_reported_input_tokens": 42,
          "cached_input_tokens": 42,
          "fresh_input_tokens": 42,
          "input_tokens": 42,
          "output_tokens": 42,
          "reasoning_output_tokens": 42,
          "reasoning_reported_output_tokens": 42
        },
        "request_attempts": 42,
        "response_count": 42,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1493664,
        "uncached_input_tokens": 39169,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11822,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11862,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11995,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12339,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 15573,
            "output_tokens": 288,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 22027,
            "output_tokens": 123,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22369,
            "output_tokens": 150,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 25513,
            "output_tokens": 178,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 27312,
            "output_tokens": 393,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28030,
            "output_tokens": 111,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28607,
            "output_tokens": 951,
            "reasoning_output_tokens": 116
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29580,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30646,
            "output_tokens": 66,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30922,
            "output_tokens": 124,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 33212,
            "output_tokens": 966,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34199,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34710,
            "output_tokens": 132,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34869,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35373,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35800,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36307,
            "output_tokens": 166,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 37157,
            "output_tokens": 62,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 39039,
            "output_tokens": 78,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39812,
            "output_tokens": 146,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 39985,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40943,
            "output_tokens": 197,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41562,
            "output_tokens": 147,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42133,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42568,
            "output_tokens": 262,
            "reasoning_output_tokens": 190
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43721,
            "output_tokens": 102,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44253,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44758,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45264,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45694,
            "output_tokens": 741,
            "reasoning_output_tokens": 269
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46456,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 46960,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47456,
            "output_tokens": 235,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47488,
            "input_tokens": 48120,
            "output_tokens": 119,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48128,
            "input_tokens": 48670,
            "output_tokens": 66,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48512,
            "input_tokens": 48876,
            "output_tokens": 69,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48768,
            "input_tokens": 49404,
            "output_tokens": 757,
            "reasoning_output_tokens": 108
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50048,
            "input_tokens": 50183,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 53,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 20,
          "bash": 1,
          "exec": 24,
          "devices": 1,
          "write": 4,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 575 rosbridge control frames; 31 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 18
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack_blocks.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/workspace/.kinex/memos/stack_blocks.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/observe_blocks.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/workspace/.kinex/tools/observe_blocks.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/stack_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-47-20261116/workspace/.kinex/tools/stack_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 98,
        "observed_images": 20
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 31,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 231,
        "published_frames": 231,
        "decode_ok": true,
        "duration_s": 5.775,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "a4e1cfd73c24120d31e95dacae3884a3bc3196ab76d9f9fd78775cd792eaa225",
        "source_sha256": {
          "third_person": "b33fab8a14cde97b9556977e1cfc9f34b72b0ae9bd5bfee7ca43ac8b9cde4bc8",
          "left_wrist": "b9ac8b936cd3ba7bb541ed795cb2213cf9f1e71815d9e3852423e754132e7ddd",
          "right_wrist": "46f54707dcdbd5af2c2da04bf53a7cac3b598b0f156c68c8cb656f35386b7983"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "1dac2482a8ee",
          "task": "robotwin/stack-blocks-two-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 232,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 231,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23000000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 232,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 231,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23000000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 232,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 231,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23000000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-48-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/48",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "48",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 28.92,
        "step_limit": 900,
        "steps": 723,
        "success": true,
        "valid": true
      },
      "steps": 723,
      "simulation_time_s": 28.92,
      "wall_time_s": 1353.405471,
      "timing": {
        "simulation_time_s": 28.92,
        "wall_time_s": 1353.405471,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Stack the small smooth brown-rimmed bowl directly over the smooth bowl with glossy finish.",
      "native_instruction": "Stack the small smooth brown-rimmed bowl directly over the smooth bowl with glossy finish.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9650168105280277,
        "cache_read_input_tokens": 1596160,
        "cache_reported_input_tokens": 1654023,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1654023,
        "cached_input_tokens": 1596160,
        "cost_usd": null,
        "fresh_input_tokens": 57863,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1654023,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1596160,
        "known_fresh_input_tokens": 57863,
        "known_input_tokens": 1654023,
        "known_output_tokens": 9512,
        "known_reasoning_output_tokens": 2777,
        "output_tokens": 9512,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2777,
        "reasoning_reported_output_tokens": 9512,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 51,
          "cache_write_input_tokens": 51,
          "cache_write_reported_input_tokens": 51,
          "cached_input_tokens": 51,
          "fresh_input_tokens": 51,
          "input_tokens": 51,
          "output_tokens": 51,
          "reasoning_output_tokens": 51,
          "reasoning_reported_output_tokens": 51
        },
        "request_attempts": 51,
        "response_count": 51,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1663535,
        "uncached_input_tokens": 57863,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11820,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11953,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12001,
            "output_tokens": 123,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12335,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18080,
            "output_tokens": 54,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18665,
            "output_tokens": 150,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 21759,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 23159,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23487,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23678,
            "output_tokens": 228,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24439,
            "output_tokens": 287,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25195,
            "output_tokens": 215,
            "reasoning_output_tokens": 175
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25876,
            "output_tokens": 396,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26526,
            "output_tokens": 980,
            "reasoning_output_tokens": 86
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27528,
            "output_tokens": 187,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28937,
            "output_tokens": 66,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29213,
            "output_tokens": 77,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 30279,
            "output_tokens": 865,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31165,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31772,
            "output_tokens": 255,
            "reasoning_output_tokens": 170
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32548,
            "output_tokens": 83,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32649,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32731,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33020,
            "output_tokens": 121,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 33663,
            "output_tokens": 272,
            "reasoning_output_tokens": 187
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34458,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 35060,
            "output_tokens": 54,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35136,
            "output_tokens": 80,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35234,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35316,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35413,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 36015,
            "output_tokens": 460,
            "reasoning_output_tokens": 231
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36705,
            "output_tokens": 611,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37838,
            "output_tokens": 110,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38477,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38525,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38607,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39217,
            "output_tokens": 157,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39898,
            "output_tokens": 143,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40566,
            "output_tokens": 51,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40635,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40721,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41334,
            "output_tokens": 574,
            "reasoning_output_tokens": 204
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42165,
            "output_tokens": 613,
            "reasoning_output_tokens": 230
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43027,
            "output_tokens": 155,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 43704,
            "output_tokens": 157,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44381,
            "output_tokens": 43,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44442,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44491,
            "output_tokens": 48,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44709,
            "output_tokens": 738,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45471,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 63,
        "model_tool_calls_by_name": {
          "read": 17,
          "ls": 4,
          "find": 1,
          "exec": 22,
          "devices": 2,
          "device": 13,
          "write": 3,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 723 rosbridge control frames; 35 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 14
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack_bowls.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/workspace/.kinex/memos/stack_bowls.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/pose_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-48-20261116/workspace/.kinex/tools/pose_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 117,
        "observed_images": 36
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 35,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 290,
        "published_frames": 290,
        "decode_ok": true,
        "duration_s": 7.25,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "7ae3776e32ffb16a5feea75412e02a054c85f1ac0a5c4853e60f20f38e935d27",
        "source_sha256": {
          "third_person": "7762944beda6181499042d76cc8d9af2ffc12db2303aac6ca346f3765150865e",
          "left_wrist": "f6072ee7edc28ce959eaedf5a6771dd4e7d6e67381ca0e1f8f569108c061b728",
          "right_wrist": "aa84bec31b77c01ae188ebb71bce50b0dcc9c2d57cf9ce347e97dbdaf6620af2"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "42944f61ec5a",
          "task": "robotwin/stack-bowls-two-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 291,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 290,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28920000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 291,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 290,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28920000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 291,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 290,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28920000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-49-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/49",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "49",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 13.16,
        "step_limit": 400,
        "steps": 329,
        "success": false,
        "valid": true
      },
      "steps": 329,
      "simulation_time_s": 13.16,
      "wall_time_s": 830.837669,
      "timing": {
        "simulation_time_s": 13.16,
        "wall_time_s": 830.837669,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Stamp Green using the block-shaped seal with angled edges",
      "native_instruction": "Stamp Green using the block-shaped seal with angled edges",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9425895474369057,
        "cache_read_input_tokens": 764672,
        "cache_reported_input_tokens": 811246,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 811246,
        "cached_input_tokens": 764672,
        "cost_usd": null,
        "fresh_input_tokens": 46574,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 811246,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 764672,
        "known_fresh_input_tokens": 46574,
        "known_input_tokens": 811246,
        "known_output_tokens": 5602,
        "known_reasoning_output_tokens": 1257,
        "output_tokens": 5602,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1257,
        "reasoning_reported_output_tokens": 5602,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 34,
          "cache_write_input_tokens": 34,
          "cache_write_reported_input_tokens": 34,
          "cached_input_tokens": 34,
          "fresh_input_tokens": 34,
          "input_tokens": 34,
          "output_tokens": 34,
          "reasoning_output_tokens": 34,
          "reasoning_reported_output_tokens": 34
        },
        "request_attempts": 34,
        "response_count": 34,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 816848,
        "uncached_input_tokens": 46574,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11809,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11844,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11975,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12023,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13467,
            "output_tokens": 188,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13568,
            "input_tokens": 16164,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 16400,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16256,
            "input_tokens": 17743,
            "output_tokens": 67,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 18341,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 18669,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 19077,
            "output_tokens": 171,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 21662,
            "output_tokens": 314,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 22199,
            "output_tokens": 898,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 23119,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 24125,
            "output_tokens": 136,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 25160,
            "output_tokens": 102,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 25472,
            "output_tokens": 87,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26254,
            "output_tokens": 76,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26573,
            "output_tokens": 810,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27404,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 28110,
            "output_tokens": 90,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28227,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28580,
            "output_tokens": 136,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29000,
            "output_tokens": 51,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29069,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29151,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29510,
            "output_tokens": 268,
            "reasoning_output_tokens": 191
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30061,
            "output_tokens": 122,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30464,
            "output_tokens": 91,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30834,
            "output_tokens": 541,
            "reasoning_output_tokens": 298
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31653,
            "output_tokens": 153,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32087,
            "output_tokens": 150,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32259,
            "output_tokens": 479,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 32761,
            "output_tokens": 17,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 39,
        "model_tool_calls_by_name": {
          "ls": 5,
          "read": 9,
          "find": 1,
          "exec": 15,
          "devices": 2,
          "write": 3,
          "edit": 1,
          "device": 2,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 329 rosbridge control frames; 36 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 8
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stamp-seal.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/workspace/.kinex/memos/stamp-seal.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/workspace/.kinex/tools/tcp_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_subgoal.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-49-20261116/workspace/.kinex/tools/tcp_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 76,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 36,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 133,
        "published_frames": 133,
        "decode_ok": true,
        "duration_s": 3.325,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "d7147d8eb61338c49f40b4386ae495441c6ca1d4bc525fcd094d6afe40c72a02",
        "source_sha256": {
          "third_person": "915e4413c04ab4709e28e6b0d6422b4196f9e4305f6157cb43e4c0a9a4835827",
          "left_wrist": "dffe1351d452ff9241a5cb54e9af07932ea69e4b224cdd9ea590f3db23d10c16",
          "right_wrist": "e746b19ed6e42a7faa120b42de5d1c7283ed4eb4e2fe7047ac9ae704582c5a0f"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "c39c34921f52",
          "task": "robotwin/stamp-seal-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 133,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 133,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13160000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 133,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 133,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13160000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 133,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 133,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13160000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": true
    },
    {
      "id": "onboard-kinex-tcp-posestate-50-20261116",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/50",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "50",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 12.64,
        "step_limit": 400,
        "steps": 316,
        "success": false,
        "valid": true
      },
      "steps": 316,
      "simulation_time_s": 12.64,
      "wall_time_s": 588.164709,
      "timing": {
        "simulation_time_s": 12.64,
        "wall_time_s": 588.164709,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Click the flat gray switch with the right arm",
      "native_instruction": "Click the flat gray switch with the right arm",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9599189179777049,
        "cache_read_input_tokens": 799360,
        "cache_reported_input_tokens": 832737,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 832737,
        "cached_input_tokens": 799360,
        "cost_usd": null,
        "fresh_input_tokens": 33377,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 832737,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 799360,
        "known_fresh_input_tokens": 33377,
        "known_input_tokens": 832737,
        "known_output_tokens": 6484,
        "known_reasoning_output_tokens": 2188,
        "output_tokens": 6484,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2188,
        "reasoning_reported_output_tokens": 6484,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 32,
          "cache_write_input_tokens": 32,
          "cache_write_reported_input_tokens": 32,
          "cached_input_tokens": 32,
          "fresh_input_tokens": 32,
          "input_tokens": 32,
          "output_tokens": 32,
          "reasoning_output_tokens": 32,
          "reasoning_reported_output_tokens": 32
        },
        "request_attempts": 32,
        "response_count": 32,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 839221,
        "uncached_input_tokens": 33377,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11814,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11854,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11987,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13253,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13312,
            "input_tokens": 17574,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 18144,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 19407,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 19571,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 19762,
            "output_tokens": 173,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 21549,
            "output_tokens": 352,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 23490,
            "output_tokens": 166,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24351,
            "output_tokens": 494,
            "reasoning_output_tokens": 217
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 24988,
            "output_tokens": 287,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25740,
            "output_tokens": 449,
            "reasoning_output_tokens": 151
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 27128,
            "output_tokens": 1045,
            "reasoning_output_tokens": 223
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28194,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28764,
            "output_tokens": 48,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28863,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29004,
            "output_tokens": 123,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29596,
            "output_tokens": 102,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30162,
            "output_tokens": 66,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30279,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30418,
            "output_tokens": 153,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 31037,
            "output_tokens": 291,
            "reasoning_output_tokens": 252
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31794,
            "output_tokens": 841,
            "reasoning_output_tokens": 759
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33105,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33657,
            "output_tokens": 152,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34276,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34815,
            "output_tokens": 317,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35788,
            "output_tokens": 142,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 35952,
            "output_tokens": 447,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36421,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 40,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 13,
          "find": 1,
          "exec": 14,
          "devices": 1,
          "device": 3,
          "write": 2,
          "relative_pose": 2,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 316 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 11-50 under the onboard-rerun protocol.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 8
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/transcript.json",
        "usage": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/usage.json",
        "analysis": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/episode.json",
        "media_check": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/timing.json",
        "native_session": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-onboard-11-50-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/turn-switch.md",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/workspace/.kinex/memos/turn-switch.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/right_tcp_move.py",
          "file": "experiments/kinex-pose-hard-onboard-11-50-20261116/onboard-kinex-tcp-posestate-50-20261116/workspace/.kinex/tools/right_tcp_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 75,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-onboard-rerun-20261115",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 128,
        "published_frames": 128,
        "decode_ok": true,
        "duration_s": 3.2,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "bb3fc35e4ec67b817cbbfa5ebc70e6e7aa1219848e3c58f999919d43c15afb80",
        "source_sha256": {
          "third_person": "9032071ddc4381a671f2c18a483f74b925f8e56576c82c6a79da7c797445848a",
          "left_wrist": "983aeccdba6de8a87deeb1639af4b86eb8c588528832ed10dcc40a7e5e50bafd",
          "right_wrist": "5f4805f07a78c8a7450852e20bd789a4b6ae0687c216c58d13d3de12d83852ea"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "8382abcfb5b6",
          "task": "robotwin/turn-switch-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 128,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 128,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 128,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 128,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 128,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 128,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-onboard-rerun-20261115",
      "campaign": "progvla-onboard-hard11-50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-jy",
      "vla_used": false
    },
    {
      "id": "kinex-pose-film-01-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/01",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "01",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 22.64,
        "step_limit": 800,
        "steps": 566,
        "success": true,
        "valid": true
      },
      "steps": 566,
      "simulation_time_s": 22.64,
      "wall_time_s": 1060.808746,
      "timing": {
        "simulation_time_s": 22.64,
        "wall_time_s": 1060.808746,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "native_instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9749557407809782,
        "cache_read_input_tokens": 1305728,
        "cache_reported_input_tokens": 1339269,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1339269,
        "cached_input_tokens": 1305728,
        "cost_usd": null,
        "fresh_input_tokens": 33541,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1339269,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1305728,
        "known_fresh_input_tokens": 33541,
        "known_input_tokens": 1339269,
        "known_output_tokens": 7602,
        "known_reasoning_output_tokens": 1572,
        "output_tokens": 7602,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1572,
        "reasoning_reported_output_tokens": 7602,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 41,
          "cache_write_input_tokens": 41,
          "cache_write_reported_input_tokens": 41,
          "cached_input_tokens": 41,
          "fresh_input_tokens": 41,
          "input_tokens": 41,
          "output_tokens": 41,
          "reasoning_output_tokens": 41,
          "reasoning_reported_output_tokens": 41
        },
        "request_attempts": 41,
        "response_count": 41,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1346871,
        "uncached_input_tokens": 33541,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11828,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11868,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12001,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12345,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18234,
            "output_tokens": 189,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 22401,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22729,
            "output_tokens": 169,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 25805,
            "output_tokens": 408,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26726,
            "output_tokens": 493,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27463,
            "output_tokens": 902,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28388,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 29331,
            "output_tokens": 61,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29454,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29706,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30210,
            "output_tokens": 164,
            "reasoning_output_tokens": 122
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31553,
            "output_tokens": 745,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32320,
            "output_tokens": 122,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32912,
            "output_tokens": 121,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33512,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 34053,
            "output_tokens": 271,
            "reasoning_output_tokens": 188
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34798,
            "output_tokens": 51,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35171,
            "output_tokens": 89,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35722,
            "output_tokens": 38,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35929,
            "output_tokens": 321,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36452,
            "output_tokens": 217,
            "reasoning_output_tokens": 140
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37149,
            "output_tokens": 69,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37280,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37641,
            "output_tokens": 192,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38314,
            "output_tokens": 76,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38871,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39431,
            "output_tokens": 245,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40377,
            "output_tokens": 125,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 41197,
            "output_tokens": 201,
            "reasoning_output_tokens": 120
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41874,
            "output_tokens": 343,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42413,
            "output_tokens": 135,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 43033,
            "output_tokens": 135,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43655,
            "output_tokens": 68,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 43865,
            "output_tokens": 58,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 43985,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44251,
            "output_tokens": 748,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45022,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 53,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 18,
          "bash": 1,
          "exec": 21,
          "devices": 4,
          "device": 2,
          "write": 3,
          "device_stop": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 566 rosbridge control frames; 40 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 14
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/handover-block.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/workspace/.kinex/memos/handover-block.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/workspace/.kinex/tools/cartesian_subgoal.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-01-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 97,
        "observed_images": 22
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 40,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 228,
        "published_frames": 228,
        "decode_ok": true,
        "duration_s": 5.7,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "711c325c28db83e7c5b0c0b76a1e5b247978b3c98b45e859a79c20228037e9c0",
        "source_sha256": {
          "third_person": "fcc31f68236b9a60319b447e5f31f13fa5c4af8672d35a58a053c47f45b71237",
          "left_wrist": "165677893de4d7198b479977c907942621145258507894991fef28d24b12284c",
          "right_wrist": "f82af06916665c365f46bb306e62a10ab6fe91f0189dac5389196331e36cc182"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "091979e654e7",
          "task": "robotwin/handover-block-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 228,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 228,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 22640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 228,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 228,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 22640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 228,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 228,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 22640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-02-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/02",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "02",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 23.52,
        "step_limit": 600,
        "steps": 588,
        "success": false,
        "valid": true
      },
      "steps": 588,
      "simulation_time_s": 23.52,
      "wall_time_s": 1235.428248,
      "timing": {
        "simulation_time_s": 23.52,
        "wall_time_s": 1235.428248,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "native_instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9403427498239298,
        "cache_read_input_tokens": 1201664,
        "cache_reported_input_tokens": 1277900,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1277900,
        "cached_input_tokens": 1201664,
        "cost_usd": null,
        "fresh_input_tokens": 76236,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1277900,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1201664,
        "known_fresh_input_tokens": 76236,
        "known_input_tokens": 1277900,
        "known_output_tokens": 6815,
        "known_reasoning_output_tokens": 1839,
        "output_tokens": 6815,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1839,
        "reasoning_reported_output_tokens": 6815,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 41,
          "cache_write_input_tokens": 41,
          "cache_write_reported_input_tokens": 41,
          "cached_input_tokens": 41,
          "fresh_input_tokens": 41,
          "input_tokens": 41,
          "output_tokens": 41,
          "reasoning_output_tokens": 41,
          "reasoning_reported_output_tokens": 41
        },
        "request_attempts": 41,
        "response_count": 41,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1284715,
        "uncached_input_tokens": 76236,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11817,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11950,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12000,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12293,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 12756,
            "output_tokens": 225,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12800,
            "input_tokens": 18188,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 18758,
            "output_tokens": 152,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 23632,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 23960,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24151,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 25230,
            "output_tokens": 175,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25991,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26399,
            "output_tokens": 306,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 28604,
            "output_tokens": 95,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29165,
            "output_tokens": 1093,
            "reasoning_output_tokens": 280
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30281,
            "output_tokens": 113,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 31071,
            "output_tokens": 140,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31889,
            "output_tokens": 71,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32170,
            "output_tokens": 60,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32292,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 33511,
            "output_tokens": 792,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34324,
            "output_tokens": 120,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34817,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35298,
            "output_tokens": 70,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35742,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 36194,
            "output_tokens": 402,
            "reasoning_output_tokens": 295
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36968,
            "output_tokens": 290,
            "reasoning_output_tokens": 214
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 37634,
            "output_tokens": 142,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38147,
            "output_tokens": 69,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 38278,
            "output_tokens": 168,
            "reasoning_output_tokens": 117
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38816,
            "output_tokens": 257,
            "reasoning_output_tokens": 138
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39450,
            "output_tokens": 121,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 39942,
            "output_tokens": 193,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 40506,
            "output_tokens": 110,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40990,
            "output_tokens": 201,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41563,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41980,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42429,
            "output_tokens": 95,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42575,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 42768,
            "output_tokens": 580,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43371,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 48,
        "model_tool_calls_by_name": {
          "read": 17,
          "ls": 1,
          "bash": 1,
          "exec": 20,
          "devices": 1,
          "device": 3,
          "write": 3,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 588 rosbridge control frames; 44 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 15
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/dual_shoes.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/workspace/.kinex/memos/dual_shoes.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-02-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 92,
        "observed_images": 23
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 44,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 236,
        "published_frames": 236,
        "decode_ok": true,
        "duration_s": 5.9,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "e01cd48e24700b442d0d6df30b1a62aa9bb29277d1e4cfc523e8407f8a9d03fd",
        "source_sha256": {
          "third_person": "ab3a11f917e89fda5a94b01dda859e5a5985c6862db1a5bd12c7dd87a1e0dac4",
          "left_wrist": "7eb6232aa30bc81a04ce8cdb4b3a6c6245687a2c7dbb0b74126fde24e8b2e224",
          "right_wrist": "a6749470be9426c3062bb2642d068731c2f770fc139d8f83ff9e80549c245f40"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ab499e5fe137",
          "task": "robotwin/place-dual-shoes-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 237,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 236,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23520000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 237,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 236,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23520000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 237,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 236,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 23520000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-03-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/03",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "03",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 2
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 43.0,
        "step_limit": 1700,
        "steps": 1075,
        "success": true,
        "valid": true
      },
      "steps": 1075,
      "simulation_time_s": 43.0,
      "wall_time_s": 1516.738348,
      "timing": {
        "simulation_time_s": 43.0,
        "wall_time_s": 1516.738348,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap in the dustbin the simple black garbage bin.",
      "native_instruction": "Place the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap in the dustbin the simple black garbage bin.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9439245916383685,
        "cache_read_input_tokens": 1332864,
        "cache_reported_input_tokens": 1412045,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1412045,
        "cached_input_tokens": 1332864,
        "cost_usd": null,
        "fresh_input_tokens": 79181,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1412045,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1332864,
        "known_fresh_input_tokens": 79181,
        "known_input_tokens": 1412045,
        "known_output_tokens": 7125,
        "known_reasoning_output_tokens": 1160,
        "output_tokens": 7125,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1160,
        "reasoning_reported_output_tokens": 7125,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 44,
          "cache_write_input_tokens": 44,
          "cache_write_reported_input_tokens": 44,
          "cached_input_tokens": 44,
          "fresh_input_tokens": 44,
          "input_tokens": 44,
          "output_tokens": 44,
          "reasoning_output_tokens": 44,
          "reasoning_reported_output_tokens": 44
        },
        "request_attempts": 44,
        "response_count": 44,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1419170,
        "uncached_input_tokens": 79181,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11837,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11877,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12010,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12060,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12353,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18242,
            "output_tokens": 187,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 22407,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22735,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 22925,
            "output_tokens": 89,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 24053,
            "output_tokens": 296,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24744,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25152,
            "output_tokens": 296,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 27220,
            "output_tokens": 52,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27593,
            "output_tokens": 1080,
            "reasoning_output_tokens": 234
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28696,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29474,
            "output_tokens": 135,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30345,
            "output_tokens": 164,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 32170,
            "output_tokens": 793,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 32984,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33407,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33803,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34329,
            "output_tokens": 278,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 34965,
            "output_tokens": 161,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35583,
            "output_tokens": 87,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36128,
            "output_tokens": 299,
            "reasoning_output_tokens": 190
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36771,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 37213,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37644,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38039,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38697,
            "output_tokens": 209,
            "reasoning_output_tokens": 96
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 39252,
            "output_tokens": 85,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39679,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40072,
            "output_tokens": 118,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 40535,
            "output_tokens": 150,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 41031,
            "output_tokens": 179,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41354,
            "output_tokens": 97,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41796,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 42197,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42605,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 43032,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43224,
            "output_tokens": 82,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43368,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 43829,
            "output_tokens": 760,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 44615,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 55,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 17,
          "bash": 1,
          "exec": 27,
          "devices": 3,
          "device": 1,
          "write": 3,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 1075 rosbridge control frames; 64 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 20
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-bottle-disposal.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/workspace/.kinex/memos/robotwin-bottle-disposal.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-03-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 102,
        "observed_images": 35
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 1700,
        "max_control_steps": 1700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 64,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 431,
        "published_frames": 431,
        "decode_ok": true,
        "duration_s": 10.775,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "de16f1b1d18b2139d3ee01ab4f83378837712cabfaf9154b1be0d9b80c743cb2",
        "source_sha256": {
          "third_person": "393d681ac8726b1ecb73357c29216425ffd7e8ebeb87faa481b71c3cf1db6d45",
          "left_wrist": "0d0d2366d1df4517a5094e4d367a3245a2acd24cada48b69dfdb67a368bd86f1",
          "right_wrist": "0da68af03983bf620b07f552dfb076ee2bdea9a1660f3a24051d393ebf415478"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "f60de2ef6a76",
          "task": "robotwin/put-bottles-dustbin-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 432,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 431,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 43000000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 432,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 431,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 43000000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 432,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 431,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 43000000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-04-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/04",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "04",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 12.96,
        "step_limit": 500,
        "steps": 324,
        "success": false,
        "valid": true
      },
      "steps": 324,
      "simulation_time_s": 12.96,
      "wall_time_s": 860.078947,
      "timing": {
        "simulation_time_s": 12.96,
        "wall_time_s": 860.078947,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "native_instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9469776431627205,
        "cache_read_input_tokens": 964224,
        "cache_reported_input_tokens": 1018212,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1018212,
        "cached_input_tokens": 964224,
        "cost_usd": null,
        "fresh_input_tokens": 53988,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1018212,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 964224,
        "known_fresh_input_tokens": 53988,
        "known_input_tokens": 1018212,
        "known_output_tokens": 7019,
        "known_reasoning_output_tokens": 2549,
        "output_tokens": 7019,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2549,
        "reasoning_reported_output_tokens": 7019,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 40,
          "cache_write_input_tokens": 40,
          "cache_write_reported_input_tokens": 40,
          "cached_input_tokens": 40,
          "fresh_input_tokens": 40,
          "input_tokens": 40,
          "output_tokens": 40,
          "reasoning_output_tokens": 40,
          "reasoning_reported_output_tokens": 40
        },
        "request_attempts": 40,
        "response_count": 40,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1025231,
        "uncached_input_tokens": 53988,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11829,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11877,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12008,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12301,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 12764,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12672,
            "input_tokens": 13988,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13824,
            "input_tokens": 14802,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14720,
            "input_tokens": 15619,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15488,
            "input_tokens": 16445,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 17273,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17152,
            "input_tokens": 18065,
            "output_tokens": 140,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 20393,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 22257,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 22727,
            "output_tokens": 38,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 22932,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23184,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 24263,
            "output_tokens": 362,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 25020,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25428,
            "output_tokens": 299,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26293,
            "output_tokens": 98,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26857,
            "output_tokens": 1121,
            "reasoning_output_tokens": 314
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 27999,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28232,
            "output_tokens": 120,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28976,
            "output_tokens": 62,
            "reasoning_output_tokens": 6
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 29065,
            "output_tokens": 120,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29550,
            "output_tokens": 154,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 30295,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30703,
            "output_tokens": 105,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31181,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31797,
            "output_tokens": 298,
            "reasoning_output_tokens": 211
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32563,
            "output_tokens": 584,
            "reasoning_output_tokens": 469
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33215,
            "output_tokens": 120,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33705,
            "output_tokens": 138,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34181,
            "output_tokens": 552,
            "reasoning_output_tokens": 330
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34971,
            "output_tokens": 504,
            "reasoning_output_tokens": 410
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 35956,
            "output_tokens": 72,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36398,
            "output_tokens": 421,
            "reasoning_output_tokens": 334
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37183,
            "output_tokens": 73,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37620,
            "output_tokens": 655,
            "reasoning_output_tokens": 204
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38297,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 42,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 16,
          "bash": 1,
          "exec": 19,
          "devices": 1,
          "device": 1,
          "write": 2,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 324 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 11
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/scan-object.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/workspace/.kinex/memos/scan-object.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_pose.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-04-20261116/workspace/.kinex/tools/dual_pose.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 85,
        "observed_images": 17
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 131,
        "published_frames": 131,
        "decode_ok": true,
        "duration_s": 3.275,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "6cf8b76a08ff1feb7b5eecb7589d774afed3d72724da72eb0594d53ac9640bee",
        "source_sha256": {
          "third_person": "337b4224cba46d88e3d6d8014d583b1f8c98c1c24610339f818190b4fe664397",
          "left_wrist": "5464a2d188956991a83a142094a12cf5a8ca2514b92f168eb8a7d0b0434dfcd2",
          "right_wrist": "ac531bef72124a146d9acb3c00d60b5a92d62c6c0a4ef85545b6c82caebcac3c"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "249c8c3c4a4f",
          "task": "robotwin/scan-object-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 131,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 131,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 131,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 131,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 131,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 131,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": false
    },
    {
      "id": "kinex-pose-film-05-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/05",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "05",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 42.96,
        "step_limit": 1200,
        "steps": 1074,
        "success": true,
        "valid": true
      },
      "steps": 1074,
      "simulation_time_s": 42.96,
      "wall_time_s": 1697.906527,
      "timing": {
        "simulation_time_s": 42.96,
        "wall_time_s": 1697.906527,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "native_instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9754935494515223,
        "cache_read_input_tokens": 1939328,
        "cache_reported_input_tokens": 1988048,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1988048,
        "cached_input_tokens": 1939328,
        "cost_usd": null,
        "fresh_input_tokens": 48720,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1988048,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1939328,
        "known_fresh_input_tokens": 48720,
        "known_input_tokens": 1988048,
        "known_output_tokens": 9885,
        "known_reasoning_output_tokens": 2404,
        "output_tokens": 9885,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2404,
        "reasoning_reported_output_tokens": 9885,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 56,
          "cache_write_input_tokens": 56,
          "cache_write_reported_input_tokens": 56,
          "cached_input_tokens": 56,
          "fresh_input_tokens": 56,
          "input_tokens": 56,
          "output_tokens": 56,
          "reasoning_output_tokens": 56,
          "reasoning_reported_output_tokens": 56
        },
        "request_attempts": 56,
        "response_count": 56,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1997933,
        "uncached_input_tokens": 48720,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11817,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11857,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11990,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12040,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13535,
            "output_tokens": 218,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13568,
            "input_tokens": 17094,
            "output_tokens": 159,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17152,
            "input_tokens": 22541,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 23940,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24268,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24676,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25225,
            "output_tokens": 354,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 26645,
            "output_tokens": 138,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27249,
            "output_tokens": 1198,
            "reasoning_output_tokens": 310
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28468,
            "output_tokens": 353,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29289,
            "output_tokens": 370,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29991,
            "output_tokens": 230,
            "reasoning_output_tokens": 151
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30720,
            "output_tokens": 142,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31359,
            "output_tokens": 75,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31644,
            "output_tokens": 85,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32768,
            "output_tokens": 914,
            "reasoning_output_tokens": 121
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33703,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34407,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34502,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34972,
            "output_tokens": 133,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35612,
            "output_tokens": 178,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35942,
            "output_tokens": 154,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36492,
            "output_tokens": 194,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37081,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37557,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37789,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38268,
            "output_tokens": 190,
            "reasoning_output_tokens": 100
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 39154,
            "output_tokens": 271,
            "reasoning_output_tokens": 180
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39824,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40305,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40494,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40975,
            "output_tokens": 302,
            "reasoning_output_tokens": 214
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41670,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42162,
            "output_tokens": 217,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42772,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43010,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 43486,
            "output_tokens": 157,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 44035,
            "output_tokens": 229,
            "reasoning_output_tokens": 140
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44656,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 44841,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45324,
            "output_tokens": 212,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45928,
            "output_tokens": 207,
            "reasoning_output_tokens": 116
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46528,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 46762,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 47544,
            "output_tokens": 139,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 48081,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48000,
            "input_tokens": 48562,
            "output_tokens": 123,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48512,
            "input_tokens": 49079,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49309,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49152,
            "input_tokens": 49402,
            "output_tokens": 51,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49280,
            "input_tokens": 49965,
            "output_tokens": 750,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50560,
            "input_tokens": 50739,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 65,
        "model_tool_calls_by_name": {
          "ls": 5,
          "read": 13,
          "find": 1,
          "exec": 40,
          "devices": 1,
          "write": 3,
          "edit": 1,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 1074 rosbridge control frames; 56 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 3,
          "direct_motion_exec_calls": 31
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack_bowls.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/workspace/.kinex/memos/stack_bowls.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/bowl_motion.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/workspace/.kinex/tools/bowl_motion.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-05-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 124,
        "observed_images": 26
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 56,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 431,
        "published_frames": 431,
        "decode_ok": true,
        "duration_s": 10.775,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "b9bd342c1fed99ce9af6732b8e30200ed2b721a2de249830f951136ec2e47908",
        "source_sha256": {
          "third_person": "427309829564c04970d2de4b54d0ce8ca39e7f6fbe1102ede5339a615e40c538",
          "left_wrist": "0676b562d933682448191abb039090f1ad2a03b3ea6657922831ead748defd0f",
          "right_wrist": "02a1c83e44a922161c1fd831278266c5f74fc7b8256fbbbfdd3b7deb3c9510b3"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "4c700f8f4056",
          "task": "robotwin/stack-bowls-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 431,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 431,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 42960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 431,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 431,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 42960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 431,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 431,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 42960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-06-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/06",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "06",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 33.480000000000004,
        "step_limit": 1200,
        "steps": 837,
        "success": true,
        "valid": true
      },
      "steps": 837,
      "simulation_time_s": 33.480000000000004,
      "wall_time_s": 1105.096567,
      "timing": {
        "simulation_time_s": 33.480000000000004,
        "wall_time_s": 1105.096567,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "native_instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.956603338555064,
        "cache_read_input_tokens": 2016384,
        "cache_reported_input_tokens": 2107858,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2107858,
        "cached_input_tokens": 2016384,
        "cost_usd": null,
        "fresh_input_tokens": 91474,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2107858,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2016384,
        "known_fresh_input_tokens": 91474,
        "known_input_tokens": 2107858,
        "known_output_tokens": 7498,
        "known_reasoning_output_tokens": 1276,
        "output_tokens": 7498,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1276,
        "reasoning_reported_output_tokens": 7498,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 66,
          "cache_write_input_tokens": 66,
          "cache_write_reported_input_tokens": 66,
          "cached_input_tokens": 66,
          "fresh_input_tokens": 66,
          "input_tokens": 66,
          "output_tokens": 66,
          "reasoning_output_tokens": 66,
          "reasoning_reported_output_tokens": 66
        },
        "request_attempts": 66,
        "response_count": 66,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2115356,
        "uncached_input_tokens": 91474,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11825,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11865,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11998,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12048,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 12341,
            "output_tokens": 206,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16609,
            "output_tokens": 82,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 17899,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18469,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 18699,
            "output_tokens": 140,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 20742,
            "output_tokens": 116,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21466,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 22545,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22953,
            "output_tokens": 412,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 24229,
            "output_tokens": 159,
            "reasoning_output_tokens": 117
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24854,
            "output_tokens": 1367,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26244,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26951,
            "output_tokens": 153,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27323,
            "output_tokens": 280,
            "reasoning_output_tokens": 165
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27632,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27919,
            "output_tokens": 87,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28152,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28398,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 28655,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28848,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29047,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29305,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 30014,
            "output_tokens": 213,
            "reasoning_output_tokens": 149
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30981,
            "output_tokens": 85,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 32245,
            "output_tokens": 97,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32561,
            "output_tokens": 127,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32877,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33131,
            "output_tokens": 96,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33378,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33627,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34450,
            "output_tokens": 87,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34566,
            "output_tokens": 65,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34820,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35073,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35277,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35530,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36353,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36607,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 36866,
            "output_tokens": 84,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37101,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37353,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37614,
            "output_tokens": 131,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 39180,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39221,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 39412,
            "output_tokens": 81,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39644,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39895,
            "output_tokens": 192,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40116,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40601,
            "output_tokens": 109,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40899,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41104,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41361,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41953,
            "output_tokens": 61,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 42449,
            "output_tokens": 154,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42822,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 43073,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43890,
            "output_tokens": 161,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 44241,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44442,
            "output_tokens": 55,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44559,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 45379,
            "output_tokens": 774,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46177,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 74,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "bash": 1,
          "exec": 44,
          "devices": 1,
          "device": 4,
          "write": 2,
          "edit": 3,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 837 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 41
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack_blocks.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/workspace/.kinex/memos/stack_blocks.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/stack_blocks.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-06-20261116/workspace/.kinex/tools/stack_blocks.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 143,
        "observed_images": 13
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 336,
        "published_frames": 336,
        "decode_ok": true,
        "duration_s": 8.4,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "1b61cf451de17cbb0af5acc70c481748958f2e69da70002d9c43f6a5460624d8",
        "source_sha256": {
          "third_person": "e242c9cf5fc1d3ef5b8d9cb5f9f8bdf026fc08976e21e15bdf122b64b4a75f00",
          "left_wrist": "5a4acfdb388919306c4fc966e415b8e89de88085882381babdf6dcf68d65b50a",
          "right_wrist": "c6d666d4c11c2f8544d20cd16288f1c74a281a71af9b7e1b7dd557fa8fea6b74"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "0c2234208b47",
          "task": "robotwin/stack-blocks-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 336,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 336,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33480000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 336,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 336,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33480000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 336,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 336,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33480000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": false
    },
    {
      "id": "kinex-pose-film-07-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/07",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "07",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 30.92,
        "step_limit": 900,
        "steps": 773,
        "success": true,
        "valid": true
      },
      "steps": 773,
      "simulation_time_s": 30.92,
      "wall_time_s": 1596.988693,
      "timing": {
        "simulation_time_s": 30.92,
        "wall_time_s": 1596.988693,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the cylindrical mug with handle on the table, rotate it, set it down in the middle, then hang it onto the rack with straight and slanted rods.",
      "native_instruction": "Grab the cylindrical mug with handle on the table, rotate it, set it down in the middle, then hang it onto the rack with straight and slanted rods.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.959298897679456,
        "cache_read_input_tokens": 2045184,
        "cache_reported_input_tokens": 2131957,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2131957,
        "cached_input_tokens": 2045184,
        "cost_usd": null,
        "fresh_input_tokens": 86773,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2131957,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2045184,
        "known_fresh_input_tokens": 86773,
        "known_input_tokens": 2131957,
        "known_output_tokens": 12722,
        "known_reasoning_output_tokens": 5690,
        "output_tokens": 12722,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 5690,
        "reasoning_reported_output_tokens": 12722,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 59,
          "cache_write_input_tokens": 59,
          "cache_write_reported_input_tokens": 59,
          "cached_input_tokens": 59,
          "fresh_input_tokens": 59,
          "input_tokens": 59,
          "output_tokens": 59,
          "reasoning_output_tokens": 59,
          "reasoning_reported_output_tokens": 59
        },
        "request_attempts": 59,
        "response_count": 59,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2144679,
        "uncached_input_tokens": 86773,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11832,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11963,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12011,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12304,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12455,
            "output_tokens": 168,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 16566,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 20382,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 20710,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 22110,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 22632,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 23040,
            "output_tokens": 125,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23588,
            "output_tokens": 318,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24279,
            "output_tokens": 1012,
            "reasoning_output_tokens": 187
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 25313,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26086,
            "output_tokens": 75,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26371,
            "output_tokens": 53,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 27413,
            "output_tokens": 907,
            "reasoning_output_tokens": 200
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28341,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28983,
            "output_tokens": 329,
            "reasoning_output_tokens": 243
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29953,
            "output_tokens": 78,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30561,
            "output_tokens": 130,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31217,
            "output_tokens": 589,
            "reasoning_output_tokens": 514
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32320,
            "output_tokens": 118,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32962,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33544,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34321,
            "output_tokens": 57,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 35552,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36072,
            "output_tokens": 107,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36206,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36925,
            "output_tokens": 365,
            "reasoning_output_tokens": 102
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37768,
            "output_tokens": 683,
            "reasoning_output_tokens": 604
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38972,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39575,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40174,
            "output_tokens": 53,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40249,
            "output_tokens": 204,
            "reasoning_output_tokens": 146
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40978,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41580,
            "output_tokens": 81,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41831,
            "output_tokens": 604,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 43154,
            "output_tokens": 280,
            "reasoning_output_tokens": 207
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43502,
            "output_tokens": 205,
            "reasoning_output_tokens": 110
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44339,
            "output_tokens": 344,
            "reasoning_output_tokens": 250
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 45318,
            "output_tokens": 489,
            "reasoning_output_tokens": 303
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 45909,
            "output_tokens": 136,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45952,
            "input_tokens": 46681,
            "output_tokens": 429,
            "reasoning_output_tokens": 154
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 47578,
            "output_tokens": 490,
            "reasoning_output_tokens": 400
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47872,
            "input_tokens": 48590,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 48782,
            "output_tokens": 270,
            "reasoning_output_tokens": 181
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 49576,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 49770,
            "output_tokens": 447,
            "reasoning_output_tokens": 198
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50048,
            "input_tokens": 50553,
            "output_tokens": 578,
            "reasoning_output_tokens": 487
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51072,
            "input_tokens": 51653,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51584,
            "input_tokens": 51845,
            "output_tokens": 473,
            "reasoning_output_tokens": 384
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52224,
            "input_tokens": 52840,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 53450,
            "output_tokens": 53,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52736,
            "input_tokens": 53673,
            "output_tokens": 72,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53504,
            "input_tokens": 53887,
            "output_tokens": 60,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53760,
            "input_tokens": 54009,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53888,
            "input_tokens": 54438,
            "output_tokens": 810,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53376,
            "input_tokens": 55271,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 66,
        "model_tool_calls_by_name": {
          "read": 16,
          "ls": 1,
          "bash": 1,
          "devices": 2,
          "exec": 34,
          "write": 3,
          "grep": 1,
          "edit": 1,
          "device_stop": 2,
          "device": 5
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 773 rosbridge control frames; 30 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 24
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/hanging-mug.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/workspace/.kinex/memos/hanging-mug.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/mug_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/workspace/.kinex/tools/mug_subgoal.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-07-20261116/workspace/.kinex/tools/tcp_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 128,
        "observed_images": 37
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 30,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 310,
        "published_frames": 310,
        "decode_ok": true,
        "duration_s": 7.75,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "428d8754e269e4040f05de2716ab88c3100635aadefafd9cb773add2de68191d",
        "source_sha256": {
          "third_person": "8f026dab463afa35407481e6f5265f5086d88b63a9605da67cdb26d0326e7990",
          "left_wrist": "7d0dbf9842dba88cfde66b784922632358814a14ca402b02178ebb944ff44583",
          "right_wrist": "436e5c0fd633befb631d766349b34a9e00a21d650478eed8e7c2115eda799451"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "eba94389cd7f",
          "task": "robotwin/hanging-mug-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 311,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 310,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 30920000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 311,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 310,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 30920000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 311,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 310,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 30920000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-08-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/08",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "08",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 27.96,
        "step_limit": 700,
        "steps": 699,
        "success": false,
        "valid": true
      },
      "steps": 699,
      "simulation_time_s": 27.96,
      "wall_time_s": 1754.644973,
      "timing": {
        "simulation_time_s": 27.96,
        "wall_time_s": 1754.644973,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Open the cabinet with smooth finish's drawer and set the blue soap with printed patterns inside with the other arm.",
      "native_instruction": "Open the cabinet with smooth finish's drawer and set the blue soap with printed patterns inside with the other arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9768379163563955,
        "cache_read_input_tokens": 2607872,
        "cache_reported_input_tokens": 2669708,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2669708,
        "cached_input_tokens": 2607872,
        "cost_usd": null,
        "fresh_input_tokens": 61836,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2669708,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2607872,
        "known_fresh_input_tokens": 61836,
        "known_input_tokens": 2669708,
        "known_output_tokens": 16550,
        "known_reasoning_output_tokens": 8179,
        "output_tokens": 16550,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 8179,
        "reasoning_reported_output_tokens": 16550,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 61,
          "cache_write_input_tokens": 61,
          "cache_write_reported_input_tokens": 61,
          "cached_input_tokens": 61,
          "fresh_input_tokens": 61,
          "input_tokens": 61,
          "output_tokens": 61,
          "reasoning_output_tokens": 61,
          "reasoning_reported_output_tokens": 61
        },
        "request_attempts": 61,
        "response_count": 61,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2686258,
        "uncached_input_tokens": 61836,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11826,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11861,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11992,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12040,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12333,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 15099,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14976,
            "input_tokens": 15459,
            "output_tokens": 312,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15616,
            "input_tokens": 22945,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23273,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23843,
            "output_tokens": 177,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24416,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24824,
            "output_tokens": 167,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 26721,
            "output_tokens": 307,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27643,
            "output_tokens": 1198,
            "reasoning_output_tokens": 308
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 28864,
            "output_tokens": 107,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 32772,
            "output_tokens": 180,
            "reasoning_output_tokens": 136
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33162,
            "output_tokens": 173,
            "reasoning_output_tokens": 131
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 34374,
            "output_tokens": 781,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35176,
            "output_tokens": 105,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35967,
            "output_tokens": 85,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 37142,
            "output_tokens": 116,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37285,
            "output_tokens": 105,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 38127,
            "output_tokens": 274,
            "reasoning_output_tokens": 144
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 39001,
            "output_tokens": 272,
            "reasoning_output_tokens": 162
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39867,
            "output_tokens": 718,
            "reasoning_output_tokens": 488
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 41392,
            "output_tokens": 180,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 42164,
            "output_tokens": 373,
            "reasoning_output_tokens": 142
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42802,
            "output_tokens": 397,
            "reasoning_output_tokens": 307
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43008,
            "input_tokens": 43936,
            "output_tokens": 106,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44779,
            "output_tokens": 156,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45524,
            "output_tokens": 110,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 46100,
            "output_tokens": 178,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46870,
            "output_tokens": 245,
            "reasoning_output_tokens": 165
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47706,
            "output_tokens": 705,
            "reasoning_output_tokens": 625
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 48465,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 49138,
            "output_tokens": 174,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49280,
            "input_tokens": 49902,
            "output_tokens": 289,
            "reasoning_output_tokens": 209
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50048,
            "input_tokens": 50783,
            "output_tokens": 927,
            "reasoning_output_tokens": 773
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51584,
            "input_tokens": 52299,
            "output_tokens": 48,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52224,
            "input_tokens": 52409,
            "output_tokens": 208,
            "reasoning_output_tokens": 167
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52480,
            "input_tokens": 52961,
            "output_tokens": 677,
            "reasoning_output_tokens": 483
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53504,
            "input_tokens": 54032,
            "output_tokens": 475,
            "reasoning_output_tokens": 156
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54400,
            "input_tokens": 55045,
            "output_tokens": 497,
            "reasoning_output_tokens": 303
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55424,
            "input_tokens": 55569,
            "output_tokens": 121,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55552,
            "input_tokens": 58322,
            "output_tokens": 297,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58496,
            "input_tokens": 58884,
            "output_tokens": 342,
            "reasoning_output_tokens": 272
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59008,
            "input_tokens": 59294,
            "output_tokens": 197,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59136,
            "input_tokens": 60085,
            "output_tokens": 398,
            "reasoning_output_tokens": 276
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60416,
            "input_tokens": 61075,
            "output_tokens": 67,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61056,
            "input_tokens": 61204,
            "output_tokens": 370,
            "reasoning_output_tokens": 270
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61440,
            "input_tokens": 62170,
            "output_tokens": 532,
            "reasoning_output_tokens": 444
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62464,
            "input_tokens": 63302,
            "output_tokens": 267,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 63488,
            "input_tokens": 63906,
            "output_tokens": 996,
            "reasoning_output_tokens": 907
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 64768,
            "input_tokens": 65491,
            "output_tokens": 121,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 65408,
            "input_tokens": 66349,
            "output_tokens": 202,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 66432,
            "input_tokens": 67145,
            "output_tokens": 210,
            "reasoning_output_tokens": 152
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 67200,
            "input_tokens": 67952,
            "output_tokens": 173,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 67968,
            "input_tokens": 68719,
            "output_tokens": 167,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 68736,
            "input_tokens": 68937,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 68864,
            "input_tokens": 69130,
            "output_tokens": 700,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 69632,
            "input_tokens": 69855,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 69,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 17,
          "bash": 2,
          "exec": 37,
          "devices": 2,
          "write": 3,
          "edit": 2,
          "device_stop": 2,
          "device": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 699 rosbridge control frames; 26 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 24
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/cabinet-manipulation.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/workspace/.kinex/memos/cabinet-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-08-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 133,
        "observed_images": 66
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 26,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 281,
        "published_frames": 281,
        "decode_ok": true,
        "duration_s": 7.025,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c06f4d50d2af73fed63523aa65544c899c510b0848eb6bbab4a3aba5c077ea1a",
        "source_sha256": {
          "third_person": "7ed21ac3e7676f6bc6fe9f9efd7859b31de9c2ea784a44e92fe7368b8d8b57d8",
          "left_wrist": "3cab06925b190a0a2c7ec93bd651143f2091c04a72a5b3baafcb1c62fde07941",
          "right_wrist": "4a7159d9813f0d9a096d9d3aa3b315395fc41587584542b57bbb07f46b79db97"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "a6329a163e08",
          "task": "robotwin/put-object-cabinet-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 281,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 281,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 281,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 281,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 281,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 281,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-09-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/09",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "09",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 20.2,
        "step_limit": 1200,
        "steps": 505,
        "success": true,
        "valid": true
      },
      "steps": 505,
      "simulation_time_s": 20.2,
      "wall_time_s": 949.231197,
      "timing": {
        "simulation_time_s": 20.2,
        "wall_time_s": 949.231197,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move large block, medium block, and small block to the center table.",
      "native_instruction": "Move large block, medium block, and small block to the center table.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9672208858696131,
        "cache_read_input_tokens": 1097728,
        "cache_reported_input_tokens": 1134930,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1134930,
        "cached_input_tokens": 1097728,
        "cost_usd": null,
        "fresh_input_tokens": 37202,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1134930,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1097728,
        "known_fresh_input_tokens": 37202,
        "known_input_tokens": 1134930,
        "known_output_tokens": 5929,
        "known_reasoning_output_tokens": 1101,
        "output_tokens": 5929,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1101,
        "reasoning_reported_output_tokens": 5929,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 38,
          "cache_write_input_tokens": 38,
          "cache_write_reported_input_tokens": 38,
          "cached_input_tokens": 38,
          "fresh_input_tokens": 38,
          "input_tokens": 38,
          "output_tokens": 38,
          "reasoning_output_tokens": 38,
          "reasoning_reported_output_tokens": 38
        },
        "request_attempts": 38,
        "response_count": 38,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1140859,
        "uncached_input_tokens": 37202,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11988,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 12038,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14066,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13952,
            "input_tokens": 14356,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 18612,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 22428,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 24069,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24397,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24587,
            "output_tokens": 107,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25063,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 26142,
            "output_tokens": 306,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 28164,
            "output_tokens": 1029,
            "reasoning_output_tokens": 247
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29218,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 30060,
            "output_tokens": 82,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30352,
            "output_tokens": 103,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31220,
            "output_tokens": 61,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31343,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 32562,
            "output_tokens": 949,
            "reasoning_output_tokens": 193
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33532,
            "output_tokens": 119,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 34048,
            "output_tokens": 189,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34629,
            "output_tokens": 104,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35106,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35610,
            "output_tokens": 179,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36186,
            "output_tokens": 186,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36768,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37203,
            "output_tokens": 103,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37699,
            "output_tokens": 147,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38239,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38702,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39130,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39581,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40036,
            "output_tokens": 111,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40534,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40730,
            "output_tokens": 82,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41072,
            "output_tokens": 696,
            "reasoning_output_tokens": 127
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41790,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 46,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "find": 1,
          "bash": 1,
          "exec": 19,
          "devices": 2,
          "device": 1,
          "write": 3,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 505 rosbridge control frames; 16 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 15
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/blocks-ranking.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/workspace/.kinex/memos/blocks-ranking.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-09-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 87,
        "observed_images": 18
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 16,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 203,
        "published_frames": 203,
        "decode_ok": true,
        "duration_s": 5.075,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "bcc409b8be7f0d26805b1d273250774857277d48da5e18413a4303eeb99b1305",
        "source_sha256": {
          "third_person": "2245d24a01a60f9900502284958125d6559a4ec375612bfa3079962777a22420",
          "left_wrist": "e3b494c32b82abc356e6743e2d62de31ab739422db1ab50cc6153de29f0ee659",
          "right_wrist": "c05869e279248ca353eab0b38a4591ebcf2ab5bde1e8f5b1a7188da8b8b46f42"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "87bb9b7d5987",
          "task": "robotwin/blocks-ranking-size-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 204,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 203,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 20200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 204,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 203,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 20200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 204,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 203,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 20200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-10-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/10",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "10",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.4,
        "step_limit": 400,
        "steps": 310,
        "success": true,
        "valid": true
      },
      "steps": 310,
      "simulation_time_s": 12.4,
      "wall_time_s": 1136.430704,
      "timing": {
        "simulation_time_s": 12.4,
        "wall_time_s": 1136.430704,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Hold the dark gray cooking kitchenpot firmly and lift it",
      "native_instruction": "Hold the dark gray cooking kitchenpot firmly and lift it",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9609905026870306,
        "cache_read_input_tokens": 1000320,
        "cache_reported_input_tokens": 1040926,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1040926,
        "cached_input_tokens": 1000320,
        "cost_usd": null,
        "fresh_input_tokens": 40606,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1040926,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1000320,
        "known_fresh_input_tokens": 40606,
        "known_input_tokens": 1040926,
        "known_output_tokens": 8282,
        "known_reasoning_output_tokens": 2427,
        "output_tokens": 8282,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2427,
        "reasoning_reported_output_tokens": 8282,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1049208,
        "uncached_input_tokens": 40606,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11814,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11854,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11987,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14420,
            "output_tokens": 234,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 17995,
            "output_tokens": 212,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 22607,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 24577,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24905,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25313,
            "output_tokens": 162,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 27207,
            "output_tokens": 317,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27940,
            "output_tokens": 140,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28546,
            "output_tokens": 594,
            "reasoning_output_tokens": 232
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29393,
            "output_tokens": 912,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30330,
            "output_tokens": 111,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 31780,
            "output_tokens": 414,
            "reasoning_output_tokens": 290
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 33535,
            "output_tokens": 70,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33815,
            "output_tokens": 1149,
            "reasoning_output_tokens": 456
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 34988,
            "output_tokens": 122,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35452,
            "output_tokens": 226,
            "reasoning_output_tokens": 121
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36067,
            "output_tokens": 51,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36180,
            "output_tokens": 226,
            "reasoning_output_tokens": 186
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 37101,
            "output_tokens": 185,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37835,
            "output_tokens": 476,
            "reasoning_output_tokens": 330
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38341,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38639,
            "output_tokens": 349,
            "reasoning_output_tokens": 148
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39479,
            "output_tokens": 205,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 41032,
            "output_tokens": 170,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41546,
            "output_tokens": 227,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42325,
            "output_tokens": 163,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42630,
            "output_tokens": 180,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43255,
            "output_tokens": 456,
            "reasoning_output_tokens": 159
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 43741,
            "output_tokens": 532,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44297,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 44,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 16,
          "find": 1,
          "exec": 16,
          "devices": 1,
          "write": 3,
          "device_stop": 1,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 310 rosbridge control frames; 53 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 3,
          "direct_motion_exec_calls": 5
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/lift-pot.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/workspace/.kinex/memos/lift-pot.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_tcp_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/workspace/.kinex/tools/dual_tcp_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-10-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 80,
        "observed_images": 17
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 53,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 125,
        "published_frames": 125,
        "decode_ok": true,
        "duration_s": 3.125,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "2935b96166ad461a43eb88fa4c90a880744e14dbff99431d20cf88fcc1a4853c",
        "source_sha256": {
          "third_person": "b1ddc90a9cd721419db4ea7b336c4989f9d806a9be7537ec26c3fe998c22dd48",
          "left_wrist": "5e2013ae460f946604011f0555c1052985036ae38af1ea30f05015ba6d9f4fde",
          "right_wrist": "4e146c6270ffb0035e73574b3f15a7d305a5cbb7ffb160f755481865ce39abac"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "a9299c5cb376",
          "task": "robotwin/lift-pot-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 126,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 125,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12400000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 126,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 125,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12400000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 126,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 125,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12400000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-11-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/11",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "11",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 2
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 14.0,
        "step_limit": 400,
        "steps": 350,
        "success": false,
        "valid": true
      },
      "steps": 350,
      "simulation_time_s": 14.0,
      "wall_time_s": 891.640184,
      "timing": {
        "simulation_time_s": 14.0,
        "wall_time_s": 891.640184,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the green plastic bottle with ridged bottom carefully using the right arm",
      "native_instruction": "Pick up the green plastic bottle with ridged bottom carefully using the right arm",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9632141243526917,
        "cache_read_input_tokens": 692864,
        "cache_reported_input_tokens": 719325,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 719325,
        "cached_input_tokens": 692864,
        "cost_usd": null,
        "fresh_input_tokens": 26461,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 719325,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 692864,
        "known_fresh_input_tokens": 26461,
        "known_input_tokens": 719325,
        "known_output_tokens": 5771,
        "known_reasoning_output_tokens": 1404,
        "output_tokens": 5771,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1404,
        "reasoning_reported_output_tokens": 5771,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 26,
          "cache_write_input_tokens": 26,
          "cache_write_reported_input_tokens": 26,
          "cached_input_tokens": 26,
          "fresh_input_tokens": 26,
          "input_tokens": 26,
          "output_tokens": 26,
          "reasoning_output_tokens": 26,
          "reasoning_reported_output_tokens": 26
        },
        "request_attempts": 26,
        "response_count": 26,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 725096,
        "uncached_input_tokens": 26461,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11822,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11862,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11995,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15585,
            "output_tokens": 199,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15616,
            "input_tokens": 19120,
            "output_tokens": 204,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 23717,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 25121,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25449,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25857,
            "output_tokens": 123,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26406,
            "output_tokens": 312,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27048,
            "output_tokens": 162,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27676,
            "output_tokens": 1135,
            "reasoning_output_tokens": 169
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28834,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 28992,
            "output_tokens": 165,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 30292,
            "output_tokens": 162,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 31594,
            "output_tokens": 206,
            "reasoning_output_tokens": 143
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31829,
            "output_tokens": 124,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 33094,
            "output_tokens": 84,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33200,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33452,
            "output_tokens": 294,
            "reasoning_output_tokens": 141
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33935,
            "output_tokens": 216,
            "reasoning_output_tokens": 174
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 35140,
            "output_tokens": 759,
            "reasoning_output_tokens": 141
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35921,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36305,
            "output_tokens": 629,
            "reasoning_output_tokens": 262
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37278,
            "output_tokens": 498,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37801,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 36,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 16,
          "find": 1,
          "exec": 10,
          "devices": 1,
          "write": 3,
          "edit": 1,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 350 rosbridge control frames; 70 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 3,
          "direct_motion_exec_calls": 2
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin_manipulation.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/workspace/.kinex/memos/robotwin_manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vertical_lift.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-11-20261116/workspace/.kinex/tools/vertical_lift.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 65,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 70,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 141,
        "published_frames": 141,
        "decode_ok": true,
        "duration_s": 3.525,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "ba9df32ec4d6a7fa6b553eb7dfa15f85063646ed0d138c57c5130dd47d8a39fe",
        "source_sha256": {
          "third_person": "8e45ddabac6675fb2a8edeaa312918fcc687211eb619bb166a901280bd14d8ef",
          "left_wrist": "92122279e0766d20feedf523cadfb3aa0a2521578e9f47f9c327f41ef4928490",
          "right_wrist": "94ab43eb18bf40f0c3b13bb137f03da4485e419b55b4309aa5dff02914250e5a"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "73d37a8098cd",
          "task": "robotwin/adjust-bottle-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 142,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 141,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14000000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 142,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 141,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14000000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 142,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 141,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14000000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-12-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/12",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "12",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 13.52,
        "step_limit": 400,
        "steps": 338,
        "success": false,
        "valid": true
      },
      "steps": 338,
      "simulation_time_s": 13.52,
      "wall_time_s": 981.127376,
      "timing": {
        "simulation_time_s": 13.52,
        "wall_time_s": 981.127376,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the hammer with black handle and hit the block",
      "native_instruction": "Grab the hammer with black handle and hit the block",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9689727981432033,
        "cache_read_input_tokens": 1442816,
        "cache_reported_input_tokens": 1489016,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1489016,
        "cached_input_tokens": 1442816,
        "cost_usd": null,
        "fresh_input_tokens": 46200,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1489016,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1442816,
        "known_fresh_input_tokens": 46200,
        "known_input_tokens": 1489016,
        "known_output_tokens": 7565,
        "known_reasoning_output_tokens": 2045,
        "output_tokens": 7565,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2045,
        "reasoning_reported_output_tokens": 7565,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 45,
          "cache_write_input_tokens": 45,
          "cache_write_reported_input_tokens": 45,
          "cached_input_tokens": 45,
          "fresh_input_tokens": 45,
          "input_tokens": 45,
          "output_tokens": 45,
          "reasoning_output_tokens": 45,
          "reasoning_reported_output_tokens": 45
        },
        "request_attempts": 45,
        "response_count": 45,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1496581,
        "uncached_input_tokens": 46200,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11814,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11854,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11987,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 16009,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 16302,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16128,
            "input_tokens": 16453,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 16624,
            "output_tokens": 168,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 20735,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 24551,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24879,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 26279,
            "output_tokens": 130,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26805,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27213,
            "output_tokens": 125,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27761,
            "output_tokens": 317,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28340,
            "output_tokens": 76,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28882,
            "output_tokens": 886,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29790,
            "output_tokens": 106,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 32732,
            "output_tokens": 64,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33006,
            "output_tokens": 757,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33784,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34500,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 35714,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35811,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36357,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36900,
            "output_tokens": 58,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37009,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37275,
            "output_tokens": 171,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37668,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38450,
            "output_tokens": 99,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 39015,
            "output_tokens": 231,
            "reasoning_output_tokens": 149
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39712,
            "output_tokens": 48,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 39928,
            "output_tokens": 228,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 41054,
            "output_tokens": 302,
            "reasoning_output_tokens": 219
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41826,
            "output_tokens": 123,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42415,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42501,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42905,
            "output_tokens": 272,
            "reasoning_output_tokens": 190
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43642,
            "output_tokens": 104,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44212,
            "output_tokens": 386,
            "reasoning_output_tokens": 127
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44750,
            "output_tokens": 603,
            "reasoning_output_tokens": 484
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45820,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 45902,
            "output_tokens": 70,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46142,
            "output_tokens": 392,
            "reasoning_output_tokens": 360
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46556,
            "output_tokens": 575,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47152,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 52,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 16,
          "find": 1,
          "bash": 2,
          "devices": 1,
          "exec": 21,
          "write": 3,
          "edit": 1,
          "device": 5,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 338 rosbridge control frames; 24 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 12
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/hammer.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/workspace/.kinex/memos/hammer.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/move_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-12-20261116/workspace/.kinex/tools/move_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 100,
        "observed_images": 17
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 24,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 136,
        "published_frames": 136,
        "decode_ok": true,
        "duration_s": 3.4,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "77877520cf271330ec0e0d24a897ffc271b250b6ce099af4c27dd4624ff1807d",
        "source_sha256": {
          "third_person": "c76cfc81a29f3bf1ba44d617f6177b6c3af3f223ed4d2162dd99c2f78c9f74b0",
          "left_wrist": "cee04f73b7bf60dd1d3ca58e597b384180e4692d17e343c696157a34b194e6ce",
          "right_wrist": "ceacf6924905fbbafc1581a9b27f37087c8ea8525f8682cad5b71e5250256dea"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "12218ee918c7",
          "task": "robotwin/beat-block-hammer-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 137,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 136,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13520000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 137,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 136,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13520000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 137,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 136,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13520000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-13-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/13",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "13",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 29.88,
        "step_limit": 1200,
        "steps": 747,
        "success": true,
        "valid": true
      },
      "steps": 747,
      "simulation_time_s": 29.88,
      "wall_time_s": 1236.944678,
      "timing": {
        "simulation_time_s": 29.88,
        "wall_time_s": 1236.944678,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set red block on the left, then green block in the center, and finally blue block on the right.",
      "native_instruction": "Set red block on the left, then green block in the center, and finally blue block on the right.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9776315070223157,
        "cache_read_input_tokens": 1555968,
        "cache_reported_input_tokens": 1591569,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1591569,
        "cached_input_tokens": 1555968,
        "cost_usd": null,
        "fresh_input_tokens": 35601,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1591569,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1555968,
        "known_fresh_input_tokens": 35601,
        "known_input_tokens": 1591569,
        "known_output_tokens": 7331,
        "known_reasoning_output_tokens": 905,
        "output_tokens": 7331,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 905,
        "reasoning_reported_output_tokens": 7331,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 50,
          "cache_write_input_tokens": 50,
          "cache_write_reported_input_tokens": 50,
          "cached_input_tokens": 50,
          "fresh_input_tokens": 50,
          "input_tokens": 50,
          "output_tokens": 50,
          "reasoning_output_tokens": 50,
          "reasoning_reported_output_tokens": 50
        },
        "request_attempts": 50,
        "response_count": 50,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1598900,
        "uncached_input_tokens": 35601,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11824,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11957,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12005,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12298,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12449,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 18193,
            "output_tokens": 59,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18783,
            "output_tokens": 134,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 22366,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22694,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22885,
            "output_tokens": 117,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23371,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 24450,
            "output_tokens": 439,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25388,
            "output_tokens": 511,
            "reasoning_output_tokens": 226
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 27321,
            "output_tokens": 815,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28158,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28813,
            "output_tokens": 69,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28944,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29196,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 30415,
            "output_tokens": 74,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30955,
            "output_tokens": 778,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31754,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32194,
            "output_tokens": 109,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32657,
            "output_tokens": 135,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33182,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33616,
            "output_tokens": 98,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34066,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34501,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34947,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35382,
            "output_tokens": 468,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35871,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36367,
            "output_tokens": 114,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36835,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37269,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37710,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38141,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38571,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39007,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39453,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39887,
            "output_tokens": 145,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40385,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40820,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41251,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41692,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42125,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42558,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42991,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43186,
            "output_tokens": 73,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43726,
            "output_tokens": 402,
            "reasoning_output_tokens": 142
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44155,
            "output_tokens": 628,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 44805,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 57,
        "model_tool_calls_by_name": {
          "read": 17,
          "ls": 1,
          "bash": 1,
          "exec": 30,
          "devices": 1,
          "device": 1,
          "write": 4,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 747 rosbridge control frames; 35 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 24
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/blocks-ranking.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/workspace/.kinex/memos/blocks-ranking.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/measure_blocks.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/workspace/.kinex/tools/measure_blocks.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/vla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-13-20261116/workspace/.kinex/tools/vla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 110,
        "observed_images": 27
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 35,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 300,
        "published_frames": 300,
        "decode_ok": true,
        "duration_s": 7.5,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "6997c70e8262013994c09a19f376f91645978054745a7688498f9e3db88ab5ef",
        "source_sha256": {
          "third_person": "aae48d286321937308cc6c842d23f0309ade71c974708bc82d89a617cf23ca59",
          "left_wrist": "da21b4604773043b038410cb34f43cf3f346b62e81ca52a69d1fec431e3c8714",
          "right_wrist": "6fd0ac6343297cb8ddc6eb504e613f11ecd4c347ae4185e76f230f23d1c80e41"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "e3b4a46d5d43",
          "task": "robotwin/blocks-ranking-rgb-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 300,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 300,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 29880000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 300,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 300,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 29880000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 300,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 300,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 29880000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-14-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/14",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "14",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 3.7600000000000002,
        "step_limit": 400,
        "steps": 94,
        "success": true,
        "valid": true
      },
      "steps": 94,
      "simulation_time_s": 3.7600000000000002,
      "wall_time_s": 308.241182,
      "timing": {
        "simulation_time_s": 3.7600000000000002,
        "wall_time_s": 308.241182,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Click the top button of the alarm-clock with beige screen",
      "native_instruction": "Click the top button of the alarm-clock with beige screen",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9673522440553559,
        "cache_read_input_tokens": 809728,
        "cache_reported_input_tokens": 837056,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 837056,
        "cached_input_tokens": 809728,
        "cost_usd": null,
        "fresh_input_tokens": 27328,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 837056,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 809728,
        "known_fresh_input_tokens": 27328,
        "known_input_tokens": 837056,
        "known_output_tokens": 4516,
        "known_reasoning_output_tokens": 699,
        "output_tokens": 4516,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 699,
        "reasoning_reported_output_tokens": 4516,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 30,
          "cache_write_input_tokens": 30,
          "cache_write_reported_input_tokens": 30,
          "cached_input_tokens": 30,
          "fresh_input_tokens": 30,
          "input_tokens": 30,
          "output_tokens": 30,
          "reasoning_output_tokens": 30,
          "reasoning_reported_output_tokens": 30
        },
        "request_attempts": 30,
        "response_count": 30,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 841572,
        "uncached_input_tokens": 27328,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11991,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15489,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15488,
            "input_tokens": 19745,
            "output_tokens": 64,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 21017,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21587,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 21751,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 21941,
            "output_tokens": 186,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 23398,
            "output_tokens": 333,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24126,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24534,
            "output_tokens": 330,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 26628,
            "output_tokens": 218,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 29960,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30060,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30417,
            "output_tokens": 246,
            "reasoning_output_tokens": 206
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31129,
            "output_tokens": 455,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32040,
            "output_tokens": 134,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32869,
            "output_tokens": 979,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 33869,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34551,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34628,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35112,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35898,
            "output_tokens": 44,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36112,
            "output_tokens": 161,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36420,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36488,
            "output_tokens": 75,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36961,
            "output_tokens": 53,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37121,
            "output_tokens": 394,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37538,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 40,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 17,
          "find": 1,
          "exec": 8,
          "devices": 4,
          "device": 2,
          "write": 2,
          "edit": 1,
          "device_stop": 1,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 94 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 4
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-click.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/workspace/.kinex/memos/robotwin-click.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-14-20261116/workspace/.kinex/tools/tcp_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 73,
        "observed_images": 7
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 39,
        "published_frames": 39,
        "decode_ok": true,
        "duration_s": 0.975,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "baf0ab290fd815134ba8ff03c5a650fa82d1abb40c953da8e64eb356b8d78ef6",
        "source_sha256": {
          "third_person": "a8c31055a2748a43f3fb2f41f86be93d2bbf6ecc7340cc9a5c738715c31be0b2",
          "left_wrist": "8874aad37f69a95ef281418f48f6781f05a1c0061fb7ddac1e348bbf9bbb3d61",
          "right_wrist": "81410b696ee03b71036661971449411798b5e25a4bb7895e362d76cabded35e2"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "949e90abbb64",
          "task": "robotwin/click-alarmclock-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 39,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 39,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3760000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 39,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 39,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3760000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 39,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 39,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3760000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": false
    },
    {
      "id": "kinex-pose-film-15-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/15",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "15",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 9.76,
        "step_limit": 400,
        "steps": 244,
        "success": false,
        "valid": true
      },
      "steps": 244,
      "simulation_time_s": 9.76,
      "wall_time_s": 594.394647,
      "timing": {
        "simulation_time_s": 9.76,
        "wall_time_s": 594.394647,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to press the medium-sized bell's top center.",
      "native_instruction": "Use the left arm to press the medium-sized bell's top center.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9648896077771301,
        "cache_read_input_tokens": 1175168,
        "cache_reported_input_tokens": 1217930,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1217930,
        "cached_input_tokens": 1175168,
        "cost_usd": null,
        "fresh_input_tokens": 42762,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1217930,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1175168,
        "known_fresh_input_tokens": 42762,
        "known_input_tokens": 1217930,
        "known_output_tokens": 8305,
        "known_reasoning_output_tokens": 2665,
        "output_tokens": 8305,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2665,
        "reasoning_reported_output_tokens": 8305,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 39,
          "cache_write_input_tokens": 39,
          "cache_write_reported_input_tokens": 39,
          "cached_input_tokens": 39,
          "fresh_input_tokens": 39,
          "input_tokens": 39,
          "output_tokens": 39,
          "reasoning_output_tokens": 39,
          "reasoning_reported_output_tokens": 39
        },
        "request_attempts": 39,
        "response_count": 39,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1226235,
        "uncached_input_tokens": 42762,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11991,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15138,
            "output_tokens": 245,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15232,
            "input_tokens": 19469,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 20731,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 21301,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21465,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 21854,
            "output_tokens": 158,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 23723,
            "output_tokens": 298,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24416,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24824,
            "output_tokens": 298,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 26136,
            "output_tokens": 206,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 27591,
            "output_tokens": 115,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28401,
            "output_tokens": 475,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29484,
            "output_tokens": 121,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30576,
            "output_tokens": 412,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31579,
            "output_tokens": 539,
            "reasoning_output_tokens": 166
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32271,
            "output_tokens": 63,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 33508,
            "output_tokens": 358,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34851,
            "output_tokens": 736,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35611,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36266,
            "output_tokens": 42,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36354,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36492,
            "output_tokens": 212,
            "reasoning_output_tokens": 112
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37235,
            "output_tokens": 75,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37436,
            "output_tokens": 142,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37689,
            "output_tokens": 136,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38596,
            "output_tokens": 58,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38672,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 38870,
            "output_tokens": 127,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39527,
            "output_tokens": 60,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39609,
            "output_tokens": 554,
            "reasoning_output_tokens": 454
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40695,
            "output_tokens": 981,
            "reasoning_output_tokens": 614
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 43167,
            "output_tokens": 275,
            "reasoning_output_tokens": 176
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43975,
            "output_tokens": 68,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44065,
            "output_tokens": 430,
            "reasoning_output_tokens": 328
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 45028,
            "output_tokens": 604,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45658,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 47,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 16,
          "find": 1,
          "exec": 15,
          "devices": 3,
          "write": 2,
          "device": 3,
          "relative_pose": 2,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 244 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 6
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-bell.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/workspace/.kinex/memos/robotwin-bell.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/left_tcp_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-15-20261116/workspace/.kinex/tools/left_tcp_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 89,
        "observed_images": 9
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 99,
        "published_frames": 99,
        "decode_ok": true,
        "duration_s": 2.475,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "8c6a495baa9458578f8092caa4a4ba6ac86e13ddacc71a5b839059d16c3a504c",
        "source_sha256": {
          "third_person": "3552d5c55191bc18d7d358edab22478e14f724b8754000ef278341338f0479ee",
          "left_wrist": "150b2176a61f65f6feae4e6dd883efefa2f9d36163ab6ac3f0861154dcdacae8",
          "right_wrist": "831243890d154590846500fdde64bf39ff2b50dcf50fbb8d7d195cf793a820e9"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "696147ecc157",
          "task": "robotwin/click-bell-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 99,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 99,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9760000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 99,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 99,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9760000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 99,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 99,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9760000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": false
    },
    {
      "id": "kinex-pose-film-16-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/16",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "16",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 19.64,
        "step_limit": 600,
        "steps": 491,
        "success": false,
        "valid": true
      },
      "steps": 491,
      "simulation_time_s": 19.64,
      "wall_time_s": 1310.539674,
      "timing": {
        "simulation_time_s": 19.64,
        "wall_time_s": 1310.539674,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the compact tabletop trashbin, pour balls into the bin.",
      "native_instruction": "Pick up the compact tabletop trashbin, pour balls into the bin.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9552307781179596,
        "cache_read_input_tokens": 1852032,
        "cache_reported_input_tokens": 1938832,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1938832,
        "cached_input_tokens": 1852032,
        "cost_usd": null,
        "fresh_input_tokens": 86800,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1938832,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1852032,
        "known_fresh_input_tokens": 86800,
        "known_input_tokens": 1938832,
        "known_output_tokens": 14519,
        "known_reasoning_output_tokens": 6796,
        "output_tokens": 14519,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 6796,
        "reasoning_reported_output_tokens": 14519,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 52,
          "cache_write_input_tokens": 52,
          "cache_write_reported_input_tokens": 52,
          "cached_input_tokens": 52,
          "fresh_input_tokens": 52,
          "input_tokens": 52,
          "output_tokens": 52,
          "reasoning_output_tokens": 52,
          "reasoning_reported_output_tokens": 52
        },
        "request_attempts": 52,
        "response_count": 52,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1953351,
        "uncached_input_tokens": 86800,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11859,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11992,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12042,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12335,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 15806,
            "output_tokens": 214,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 20427,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 21826,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 22154,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22346,
            "output_tokens": 190,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22931,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23339,
            "output_tokens": 137,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 24122,
            "output_tokens": 281,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 25128,
            "output_tokens": 126,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 25720,
            "output_tokens": 954,
            "reasoning_output_tokens": 172
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26699,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 31546,
            "output_tokens": 60,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31668,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 31920,
            "output_tokens": 137,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 33236,
            "output_tokens": 803,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34062,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34863,
            "output_tokens": 464,
            "reasoning_output_tokens": 252
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 35802,
            "output_tokens": 273,
            "reasoning_output_tokens": 184
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 36795,
            "output_tokens": 216,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37404,
            "output_tokens": 139,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38260,
            "output_tokens": 248,
            "reasoning_output_tokens": 207
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 39203,
            "output_tokens": 1110,
            "reasoning_output_tokens": 1026
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 41034,
            "output_tokens": 221,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41469,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 42266,
            "output_tokens": 258,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42859,
            "output_tokens": 420,
            "reasoning_output_tokens": 321
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43994,
            "output_tokens": 117,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44173,
            "output_tokens": 158,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 44807,
            "output_tokens": 448,
            "reasoning_output_tokens": 242
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 45726,
            "output_tokens": 505,
            "reasoning_output_tokens": 408
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46949,
            "output_tokens": 179,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47394,
            "output_tokens": 248,
            "reasoning_output_tokens": 166
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 48346,
            "output_tokens": 241,
            "reasoning_output_tokens": 216
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48512,
            "input_tokens": 48756,
            "output_tokens": 756,
            "reasoning_output_tokens": 442
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49408,
            "input_tokens": 49905,
            "output_tokens": 348,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50176,
            "input_tokens": 50646,
            "output_tokens": 127,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50688,
            "input_tokens": 50813,
            "output_tokens": 376,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51072,
            "input_tokens": 51347,
            "output_tokens": 603,
            "reasoning_output_tokens": 511
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51840,
            "input_tokens": 52666,
            "output_tokens": 349,
            "reasoning_output_tokens": 182
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52864,
            "input_tokens": 53284,
            "output_tokens": 367,
            "reasoning_output_tokens": 279
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53504,
            "input_tokens": 54367,
            "output_tokens": 431,
            "reasoning_output_tokens": 193
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54656,
            "input_tokens": 55247,
            "output_tokens": 889,
            "reasoning_output_tokens": 798
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55936,
            "input_tokens": 56850,
            "output_tokens": 109,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56832,
            "input_tokens": 57021,
            "output_tokens": 484,
            "reasoning_output_tokens": 223
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57472,
            "input_tokens": 57536,
            "output_tokens": 223,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57728,
            "input_tokens": 57788,
            "output_tokens": 473,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58112,
            "input_tokens": 58285,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 60,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 17,
          "bash": 1,
          "exec": 27,
          "devices": 2,
          "device": 2,
          "write": 3,
          "device_stop": 3,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 491 rosbridge control frames; 20 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 15
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/dump-bin.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/workspace/.kinex/memos/dump-bin.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-16-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 115,
        "observed_images": 37
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 20,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 198,
        "published_frames": 198,
        "decode_ok": true,
        "duration_s": 4.95,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "15608f1b2cfa3c276e69c44b9ddc41c2551e197428a5c4e7a21562a069fa5d26",
        "source_sha256": {
          "third_person": "744e5e3f20a770179c2be613b74ae9229dde42afb3a6378dbf0d85118d008d2d",
          "left_wrist": "2f2143767c8032217caa6e421a0079fe2d44b06b76555b6f328004e80cb21434",
          "right_wrist": "75263a0d69173c6e8af0578be65a21319e5ca72a51bd4ad03ed23a6f9dec9981"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "9041bffc3fcf",
          "task": "robotwin/dump-bin-bigbin-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 198,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 198,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 198,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 198,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 198,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 198,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-17-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/17",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "17",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 15.72,
        "step_limit": 400,
        "steps": 393,
        "success": true,
        "valid": true
      },
      "steps": 393,
      "simulation_time_s": 15.72,
      "wall_time_s": 1324.577451,
      "timing": {
        "simulation_time_s": 15.72,
        "wall_time_s": 1324.577451,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use both arms to grab the light brown wooden roller.",
      "native_instruction": "Use both arms to grab the light brown wooden roller.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9557640633013132,
        "cache_read_input_tokens": 762240,
        "cache_reported_input_tokens": 797519,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 797519,
        "cached_input_tokens": 762240,
        "cost_usd": null,
        "fresh_input_tokens": 35279,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 797519,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 762240,
        "known_fresh_input_tokens": 35279,
        "known_input_tokens": 797519,
        "known_output_tokens": 5600,
        "known_reasoning_output_tokens": 1117,
        "output_tokens": 5600,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1117,
        "reasoning_reported_output_tokens": 5600,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 29,
          "cache_write_input_tokens": 29,
          "cache_write_reported_input_tokens": 29,
          "cached_input_tokens": 29,
          "fresh_input_tokens": 29,
          "input_tokens": 29,
          "output_tokens": 29,
          "reasoning_output_tokens": 29,
          "reasoning_reported_output_tokens": 29
        },
        "request_attempts": 29,
        "response_count": 29,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 803119,
        "uncached_input_tokens": 35279,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11814,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11854,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11987,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12035,
            "output_tokens": 134,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12383,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18127,
            "output_tokens": 179,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 21958,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22286,
            "output_tokens": 163,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 25356,
            "output_tokens": 303,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 26202,
            "output_tokens": 83,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26751,
            "output_tokens": 1006,
            "reasoning_output_tokens": 156
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 27779,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 27937,
            "output_tokens": 114,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 29198,
            "output_tokens": 301,
            "reasoning_output_tokens": 171
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30649,
            "output_tokens": 207,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 32007,
            "output_tokens": 186,
            "reasoning_output_tokens": 142
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32403,
            "output_tokens": 54,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32479,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32536,
            "output_tokens": 217,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32982,
            "output_tokens": 768,
            "reasoning_output_tokens": 148
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33772,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34193,
            "output_tokens": 119,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34683,
            "output_tokens": 173,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 34918,
            "output_tokens": 171,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35460,
            "output_tokens": 79,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35913,
            "output_tokens": 147,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36202,
            "output_tokens": 64,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36556,
            "output_tokens": 520,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37099,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 44,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "find": 4,
          "exec": 13,
          "devices": 2,
          "device": 1,
          "write": 3,
          "device_stop": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 393 rosbridge control frames; 83 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 3,
          "direct_motion_exec_calls": 5
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/grab_roller.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/workspace/.kinex/memos/grab_roller.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_tcp_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/workspace/.kinex/tools/dual_tcp_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/roller_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-17-20261116/workspace/.kinex/tools/roller_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 76,
        "observed_images": 15
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 83,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 158,
        "published_frames": 158,
        "decode_ok": true,
        "duration_s": 3.95,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "83cd7cf3394101dbec538318d64bfcd3f1f1149c5bbadda98aa46cd88b60ba0a",
        "source_sha256": {
          "third_person": "4b0bf439727b3635ee978ae7cd1ce4f5f6cb4aa6f6eedddcc8f898a654b6b2b3",
          "left_wrist": "6bd44f14430502a05b9be4027e754f7bf9ead4c596af5f0684e06d73a6344e85",
          "right_wrist": "c710460442c0161ac123a9d18b339b660a9f1bda329e9bcba7c90923adb67e3a"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "9e2dc52fa555",
          "task": "robotwin/grab-roller-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 159,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 158,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15720000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 159,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 158,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15720000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 159,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 158,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15720000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-18-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/18",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "18",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 16.88,
        "step_limit": 600,
        "steps": 422,
        "success": true,
        "valid": true
      },
      "steps": 422,
      "simulation_time_s": 16.88,
      "wall_time_s": 914.386707,
      "timing": {
        "simulation_time_s": 16.88,
        "wall_time_s": 914.386707,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the microphone used for audio input and pass it to the other arm.",
      "native_instruction": "Grab the microphone used for audio input and pass it to the other arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9715133702823594,
        "cache_read_input_tokens": 1108352,
        "cache_reported_input_tokens": 1140851,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1140851,
        "cached_input_tokens": 1108352,
        "cost_usd": null,
        "fresh_input_tokens": 32499,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1140851,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1108352,
        "known_fresh_input_tokens": 32499,
        "known_input_tokens": 1140851,
        "known_output_tokens": 6606,
        "known_reasoning_output_tokens": 1383,
        "output_tokens": 6606,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1383,
        "reasoning_reported_output_tokens": 6606,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 39,
          "cache_write_input_tokens": 39,
          "cache_write_reported_input_tokens": 39,
          "cached_input_tokens": 39,
          "fresh_input_tokens": 39,
          "input_tokens": 39,
          "output_tokens": 39,
          "reasoning_output_tokens": 39,
          "reasoning_reported_output_tokens": 39
        },
        "request_attempts": 39,
        "response_count": 39,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1147457,
        "uncached_input_tokens": 32499,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11817,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11857,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11990,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12040,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12333,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12484,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 12631,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12672,
            "input_tokens": 18375,
            "output_tokens": 178,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 22205,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22533,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22724,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23132,
            "output_tokens": 236,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 25115,
            "output_tokens": 1052,
            "reasoning_output_tokens": 158
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26189,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26305,
            "output_tokens": 191,
            "reasoning_output_tokens": 110
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 27721,
            "output_tokens": 119,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 29037,
            "output_tokens": 860,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 29918,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30487,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 31051,
            "output_tokens": 72,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31613,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32185,
            "output_tokens": 587,
            "reasoning_output_tokens": 477
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33514,
            "output_tokens": 56,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33632,
            "output_tokens": 60,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 35652,
            "output_tokens": 135,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35814,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36764,
            "output_tokens": 181,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37437,
            "output_tokens": 125,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 38047,
            "output_tokens": 267,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38523,
            "output_tokens": 184,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39199,
            "output_tokens": 56,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39317,
            "output_tokens": 205,
            "reasoning_output_tokens": 112
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 40010,
            "output_tokens": 77,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40579,
            "output_tokens": 124,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 41192,
            "output_tokens": 113,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41447,
            "output_tokens": 43,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41552,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41890,
            "output_tokens": 625,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42368,
            "input_tokens": 42540,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 48,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "bash": 2,
          "exec": 18,
          "devices": 2,
          "device": 1,
          "write": 3,
          "device_stop": 3,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 422 rosbridge control frames; 35 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 13
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/handover-microphone.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/workspace/.kinex/memos/handover-microphone.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-18-20261116/workspace/.kinex/tools/tcp_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 90,
        "observed_images": 15
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 35,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 170,
        "published_frames": 170,
        "decode_ok": true,
        "duration_s": 4.25,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "6d1c478d1f7d46af040c6fbd02a2977526f38cbbbd401db22dda4609c6c8aa9a",
        "source_sha256": {
          "third_person": "54c167c8019940443bfb194df026956037e0ee1647fa6903175ce09fcb672c95",
          "left_wrist": "30a54eaccf2d6dc35184620ea5bfddd5355b8d97f0119ed4a7d0fcbd09adfe66",
          "right_wrist": "cc2e4fd0a47a4501864e81565e6ea7b33a0ae4876b289543bd9b56036c899818"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "cba4132e5fa6",
          "task": "robotwin/handover-mic-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 170,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 170,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16880000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 170,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 170,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16880000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 170,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 170,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16880000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-19-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/19",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "19",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 10.72,
        "step_limit": 400,
        "steps": 268,
        "success": true,
        "valid": true
      },
      "steps": 268,
      "simulation_time_s": 10.72,
      "wall_time_s": 738.500303,
      "timing": {
        "simulation_time_s": 10.72,
        "wall_time_s": 738.500303,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the medium sauce can and place it near the smooth gray pot with matte finish",
      "native_instruction": "Grab the medium sauce can and place it near the smooth gray pot with matte finish",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9657409494702445,
        "cache_read_input_tokens": 968192,
        "cache_reported_input_tokens": 1002538,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1002538,
        "cached_input_tokens": 968192,
        "cost_usd": null,
        "fresh_input_tokens": 34346,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1002538,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 968192,
        "known_fresh_input_tokens": 34346,
        "known_input_tokens": 1002538,
        "known_output_tokens": 6173,
        "known_reasoning_output_tokens": 991,
        "output_tokens": 6173,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 991,
        "reasoning_reported_output_tokens": 6173,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 31,
          "cache_write_input_tokens": 31,
          "cache_write_reported_input_tokens": 31,
          "cached_input_tokens": 31,
          "fresh_input_tokens": 31,
          "input_tokens": 31,
          "output_tokens": 31,
          "reasoning_output_tokens": 31,
          "reasoning_reported_output_tokens": 31
        },
        "request_attempts": 31,
        "response_count": 31,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1008711,
        "uncached_input_tokens": 34346,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11820,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11860,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11993,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15595,
            "output_tokens": 218,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15616,
            "input_tokens": 19154,
            "output_tokens": 202,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 23756,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24084,
            "output_tokens": 152,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24631,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25039,
            "output_tokens": 153,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26923,
            "output_tokens": 301,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27803,
            "output_tokens": 397,
            "reasoning_output_tokens": 161
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28456,
            "output_tokens": 945,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29424,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 32124,
            "output_tokens": 126,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 33447,
            "output_tokens": 123,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 35822,
            "output_tokens": 1061,
            "reasoning_output_tokens": 164
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36904,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37665,
            "output_tokens": 59,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38898,
            "output_tokens": 100,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39299,
            "output_tokens": 93,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39898,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40468,
            "output_tokens": 113,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40756,
            "output_tokens": 194,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41178,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41943,
            "output_tokens": 95,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42539,
            "output_tokens": 192,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43235,
            "output_tokens": 136,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43877,
            "output_tokens": 169,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44188,
            "output_tokens": 156,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44551,
            "output_tokens": 632,
            "reasoning_output_tokens": 104
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45208,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 43,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 16,
          "find": 1,
          "devices": 1,
          "exec": 18,
          "write": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 268 rosbridge control frames; 18 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 10
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin_manipulation.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/workspace/.kinex/memos/robotwin_manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-19-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 77,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 18,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 108,
        "published_frames": 108,
        "decode_ok": true,
        "duration_s": 2.7,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "e04c02a5e3b29493163d25fac6d962e8e7e81edf0a40d2a52ecb6488df991969",
        "source_sha256": {
          "third_person": "f1381c7495d4f8e25ad29305689f2cdaf26782722e0d94ee8953a903907673a5",
          "left_wrist": "eedcdb9b557ece124ab279e9df3c501b1119660a13afca3da2d65ef3faddff8c",
          "right_wrist": "3096143d3a3543807e33c536c4c59902a7c2b36779c130114b13f88058fbb3ce"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "8a8272b9f537",
          "task": "robotwin/move-can-pot-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 109,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 108,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10720000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 109,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 108,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10720000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 109,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 108,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10720000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-20-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/20",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "20",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.64,
        "step_limit": 400,
        "steps": 316,
        "success": true,
        "valid": true
      },
      "steps": 316,
      "simulation_time_s": 12.64,
      "wall_time_s": 939.091904,
      "timing": {
        "simulation_time_s": 12.64,
        "wall_time_s": 939.091904,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the medium-sized bottle and put it onto the pad",
      "native_instruction": "Pick the medium-sized bottle and put it onto the pad",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.974789628753883,
        "cache_read_input_tokens": 1254912,
        "cache_reported_input_tokens": 1287367,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1287367,
        "cached_input_tokens": 1254912,
        "cost_usd": null,
        "fresh_input_tokens": 32455,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1287367,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1254912,
        "known_fresh_input_tokens": 32455,
        "known_input_tokens": 1287367,
        "known_output_tokens": 7887,
        "known_reasoning_output_tokens": 1823,
        "output_tokens": 7887,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1823,
        "reasoning_reported_output_tokens": 7887,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 40,
          "cache_write_input_tokens": 40,
          "cache_write_reported_input_tokens": 40,
          "cached_input_tokens": 40,
          "fresh_input_tokens": 40,
          "input_tokens": 40,
          "output_tokens": 40,
          "reasoning_output_tokens": 40,
          "reasoning_reported_output_tokens": 40
        },
        "request_attempts": 40,
        "response_count": 40,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1295254,
        "uncached_input_tokens": 32455,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11988,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14460,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 18716,
            "output_tokens": 221,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 22604,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 24245,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24573,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24763,
            "output_tokens": 60,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25142,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 26221,
            "output_tokens": 319,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26935,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27343,
            "output_tokens": 319,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 29386,
            "output_tokens": 1160,
            "reasoning_output_tokens": 305
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30569,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 31179,
            "output_tokens": 86,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31475,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32504,
            "output_tokens": 386,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33465,
            "output_tokens": 108,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34039,
            "output_tokens": 745,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34805,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35225,
            "output_tokens": 157,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35707,
            "output_tokens": 187,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36194,
            "output_tokens": 110,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36644,
            "output_tokens": 212,
            "reasoning_output_tokens": 133
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37190,
            "output_tokens": 164,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37638,
            "output_tokens": 140,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37800,
            "output_tokens": 216,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38310,
            "output_tokens": 115,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 39120,
            "output_tokens": 83,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 40553,
            "output_tokens": 510,
            "reasoning_output_tokens": 291
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41344,
            "output_tokens": 260,
            "reasoning_output_tokens": 201
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41631,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 42075,
            "output_tokens": 110,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42518,
            "output_tokens": 212,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 42872,
            "output_tokens": 44,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43085,
            "output_tokens": 151,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43344,
            "output_tokens": 345,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 43716,
            "output_tokens": 578,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44319,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 50,
        "model_tool_calls_by_name": {
          "ls": 3,
          "read": 18,
          "find": 1,
          "exec": 16,
          "devices": 3,
          "device": 2,
          "write": 3,
          "device_stop": 1,
          "grep": 1,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 316 rosbridge control frames; 25 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 8
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/aloha_bottle_pad.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/workspace/.kinex/memos/aloha_bottle_pad.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-20-20261116/workspace/.kinex/tools/tcp_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 93,
        "observed_images": 13
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 25,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 128,
        "published_frames": 128,
        "decode_ok": true,
        "duration_s": 3.2,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "425a2e82c7cf29f29edcabbcb99631fcb5a4d31c4baddb56615a2608567f7603",
        "source_sha256": {
          "third_person": "d18ee5eaa27d813880f4af3c9caf52b4076f36ae19da11eaabf7e36f014a9ed0",
          "left_wrist": "9e2551a30c0c44f789578023c459195ba1ecc9d506a717363927fa0597541fcd",
          "right_wrist": "2122fa514788827da49c098f31acfabed36417c0c718da9ca4372f2e2bddbde7"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "6c8a261a1c80",
          "task": "robotwin/move-pillbottle-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 128,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 128,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 128,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 128,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 128,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 128,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-21-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/21",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "21",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 6.2,
        "step_limit": 400,
        "steps": 155,
        "success": false,
        "valid": true
      },
      "steps": 155,
      "simulation_time_s": 6.2,
      "wall_time_s": 346.101318,
      "timing": {
        "simulation_time_s": 6.2,
        "wall_time_s": 346.101318,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the small packet of playingcards and move it outward.",
      "native_instruction": "Pick up the small packet of playingcards and move it outward.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.901871943008051,
        "cache_read_input_tokens": 773504,
        "cache_reported_input_tokens": 857665,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 857665,
        "cached_input_tokens": 773504,
        "cost_usd": null,
        "fresh_input_tokens": 84161,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 857665,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 773504,
        "known_fresh_input_tokens": 84161,
        "known_input_tokens": 857665,
        "known_output_tokens": 4751,
        "known_reasoning_output_tokens": 955,
        "output_tokens": 4751,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 955,
        "reasoning_reported_output_tokens": 4751,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 24,
          "cache_write_input_tokens": 24,
          "cache_write_reported_input_tokens": 24,
          "cached_input_tokens": 24,
          "fresh_input_tokens": 24,
          "input_tokens": 24,
          "output_tokens": 24,
          "reasoning_output_tokens": 24,
          "reasoning_reported_output_tokens": 24
        },
        "request_attempts": 24,
        "response_count": 24,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 862416,
        "uncached_input_tokens": 84161,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11988,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12332,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16588,
            "output_tokens": 170,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 19327,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 19557,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 19747,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 19999,
            "output_tokens": 199,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20096,
            "input_tokens": 42085,
            "output_tokens": 245,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 43366,
            "output_tokens": 141,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 44686,
            "output_tokens": 408,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45459,
            "output_tokens": 1242,
            "reasoning_output_tokens": 215
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 46724,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 46884,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47672,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 47782,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 48069,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48000,
            "input_tokens": 48655,
            "output_tokens": 137,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48640,
            "input_tokens": 49286,
            "output_tokens": 299,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49408,
            "input_tokens": 49931,
            "output_tokens": 115,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 50556,
            "output_tokens": 233,
            "reasoning_output_tokens": 143
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49920,
            "input_tokens": 51294,
            "output_tokens": 688,
            "reasoning_output_tokens": 239
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50688,
            "input_tokens": 52008,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 31,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 13,
          "bash": 1,
          "devices": 1,
          "device": 1,
          "exec": 11,
          "write": 2,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 155 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 7
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/aloha_packet_manipulation.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/workspace/.kinex/memos/aloha_packet_manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-21-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 58,
        "observed_images": 6
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 63,
        "published_frames": 63,
        "decode_ok": true,
        "duration_s": 1.575,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "0b36b7595e3263db12620c3eb234a97382e03be2869a265d8c01c3ad3f0b2a3a",
        "source_sha256": {
          "third_person": "5170480152df0bf3856829f38c66bd8b0ccf3fc58a470054112bc746b2aeb2cf",
          "left_wrist": "3e2a38d965457c561f78655dcd9ec49ee0a4d23ae5f7e158ade0cd73897caf5b",
          "right_wrist": "dbd5c1a63cc0b7c275c75c864d8db40091bb71a9647b60d07bb47fa6b535dc25"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "508fc68358f4",
          "task": "robotwin/move-playingcard-away-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 64,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 63,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 6200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 64,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 63,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 6200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 64,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 63,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 6200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": false
    },
    {
      "id": "kinex-pose-film-22-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/22",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "22",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 13.68,
        "step_limit": 400,
        "steps": 342,
        "success": false,
        "valid": true
      },
      "steps": 342,
      "simulation_time_s": 13.68,
      "wall_time_s": 736.294715,
      "timing": {
        "simulation_time_s": 13.68,
        "wall_time_s": 736.294715,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the red plastic stapler and set it on Red mat",
      "native_instruction": "Grab the red plastic stapler and set it on Red mat",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.959359538418707,
        "cache_read_input_tokens": 875264,
        "cache_reported_input_tokens": 912342,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 912342,
        "cached_input_tokens": 875264,
        "cost_usd": null,
        "fresh_input_tokens": 37078,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 912342,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 875264,
        "known_fresh_input_tokens": 37078,
        "known_input_tokens": 912342,
        "known_output_tokens": 5959,
        "known_reasoning_output_tokens": 698,
        "output_tokens": 5959,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 698,
        "reasoning_reported_output_tokens": 5959,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 918301,
        "uncached_input_tokens": 37078,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11814,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11854,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11987,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12037,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13234,
            "output_tokens": 206,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13312,
            "input_tokens": 17502,
            "output_tokens": 181,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 21326,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21654,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 23054,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23462,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24010,
            "output_tokens": 107,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24435,
            "output_tokens": 479,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25205,
            "output_tokens": 63,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25734,
            "output_tokens": 905,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26664,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26826,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 30833,
            "output_tokens": 51,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30946,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31198,
            "output_tokens": 137,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 32514,
            "output_tokens": 869,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33406,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33796,
            "output_tokens": 114,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34206,
            "output_tokens": 149,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34651,
            "output_tokens": 154,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35004,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35401,
            "output_tokens": 156,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35854,
            "output_tokens": 157,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36307,
            "output_tokens": 153,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36660,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37050,
            "output_tokens": 385,
            "reasoning_output_tokens": 108
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37466,
            "output_tokens": 184,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37849,
            "output_tokens": 527,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38403,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 41,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "find": 1,
          "devices": 2,
          "exec": 15,
          "write": 3,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 342 rosbridge control frames; 20 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 9
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-22-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 77,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 20,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 138,
        "published_frames": 138,
        "decode_ok": true,
        "duration_s": 3.45,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "82d6902ec45a83ecb8e4ad111310394f7104dace39314152c721dd63eeae6db8",
        "source_sha256": {
          "third_person": "6eda0f6e8dd3a7c2b6ff3ca0524d75343ddd9f522ae7f85d3d3978ca5b3b9d6a",
          "left_wrist": "eacf9875b40f2c5ac9d702d8ef9a3fb2e51b256706130385c5c9612248d0de42",
          "right_wrist": "342e2db2ad80634dab594c67560070c70892c56381b5a694d120a3d7ad2ca9b2"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "6253a92a6cb3",
          "task": "robotwin/move-stapler-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 138,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13680000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 138,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13680000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 138,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13680000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-23-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/23",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "23",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 4.8,
        "step_limit": 700,
        "steps": 120,
        "success": true,
        "valid": true
      },
      "steps": 120,
      "simulation_time_s": 4.8,
      "wall_time_s": 608.702101,
      "timing": {
        "simulation_time_s": 4.8,
        "wall_time_s": 608.702101,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift the black laptop using the right arm.",
      "native_instruction": "Lift the black laptop using the right arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9320402370175146,
        "cache_read_input_tokens": 413056,
        "cache_reported_input_tokens": 443174,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 443174,
        "cached_input_tokens": 413056,
        "cost_usd": null,
        "fresh_input_tokens": 30118,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 443174,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 413056,
        "known_fresh_input_tokens": 30118,
        "known_input_tokens": 443174,
        "known_output_tokens": 3896,
        "known_reasoning_output_tokens": 639,
        "output_tokens": 3896,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 639,
        "reasoning_reported_output_tokens": 3896,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 19,
          "cache_write_input_tokens": 19,
          "cache_write_reported_input_tokens": 19,
          "cached_input_tokens": 19,
          "fresh_input_tokens": 19,
          "input_tokens": 19,
          "output_tokens": 19,
          "reasoning_output_tokens": 19,
          "reasoning_reported_output_tokens": 19
        },
        "request_attempts": 19,
        "response_count": 19,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 447070,
        "uncached_input_tokens": 30118,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11811,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11944,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11994,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12287,
            "output_tokens": 192,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 16548,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 18235,
            "output_tokens": 175,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 22046,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22374,
            "output_tokens": 168,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 25448,
            "output_tokens": 308,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26446,
            "output_tokens": 358,
            "reasoning_output_tokens": 141
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27237,
            "output_tokens": 150,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27853,
            "output_tokens": 1035,
            "reasoning_output_tokens": 187
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28911,
            "output_tokens": 124,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29100,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29432,
            "output_tokens": 53,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29547,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30196,
            "output_tokens": 459,
            "reasoning_output_tokens": 127
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30684,
            "output_tokens": 376,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31081,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 30,
        "model_tool_calls_by_name": {
          "read": 15,
          "ls": 1,
          "bash": 1,
          "exec": 6,
          "devices": 2,
          "device": 1,
          "write": 2,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 120 rosbridge control frames; 31 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 0
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/laptop.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/workspace/.kinex/memos/laptop.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-23-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 52,
        "observed_images": 3
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 31,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 49,
        "published_frames": 49,
        "decode_ok": true,
        "duration_s": 1.225,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "9958981724eaf42218fe0ddca641a8f477c0cc77d92342c22ed94b46e0870d6c",
        "source_sha256": {
          "third_person": "38a6a0b769f49b7627c2791acc57d46aa31b538f50a2a5f3ebb16dcf9178998c",
          "left_wrist": "2bf0c74750c4c81f567f937c537e240c955014f549cc67f67da1d4a27794d0fd",
          "right_wrist": "b943f209678c18a3972e486cd3733e5be7c38ad0ce1bdaf216167fab58a8a901"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "f2799a75dec4",
          "task": "robotwin/open-laptop-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 50,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 49,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 4800000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 50,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 49,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 4800000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 50,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 49,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 4800000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-24-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/24",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "24",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 28.44,
        "step_limit": 1500,
        "steps": 711,
        "success": true,
        "valid": true
      },
      "steps": 711,
      "simulation_time_s": 28.44,
      "wall_time_s": 1208.043306,
      "timing": {
        "simulation_time_s": 28.44,
        "wall_time_s": 1208.043306,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to pull the handle of the kitchen microwave with slanted shape.",
      "native_instruction": "Use the left arm to pull the handle of the kitchen microwave with slanted shape.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9707780818857175,
        "cache_read_input_tokens": 868992,
        "cache_reported_input_tokens": 895150,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 895150,
        "cached_input_tokens": 868992,
        "cost_usd": null,
        "fresh_input_tokens": 26158,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 895150,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 868992,
        "known_fresh_input_tokens": 26158,
        "known_input_tokens": 895150,
        "known_output_tokens": 6512,
        "known_reasoning_output_tokens": 1288,
        "output_tokens": 6512,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1288,
        "reasoning_reported_output_tokens": 6512,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 901662,
        "uncached_input_tokens": 26158,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11951,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11999,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12292,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 16548,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 20364,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20692,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 22092,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22450,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23529,
            "output_tokens": 262,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24186,
            "output_tokens": 263,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24954,
            "output_tokens": 969,
            "reasoning_output_tokens": 154
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25946,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 27057,
            "output_tokens": 47,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27166,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27418,
            "output_tokens": 99,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28212,
            "output_tokens": 294,
            "reasoning_output_tokens": 145
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28815,
            "output_tokens": 103,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 30097,
            "output_tokens": 1061,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31179,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31628,
            "output_tokens": 115,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32115,
            "output_tokens": 144,
            "reasoning_output_tokens": 104
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32725,
            "output_tokens": 132,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33230,
            "output_tokens": 75,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33327,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33519,
            "output_tokens": 210,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34023,
            "output_tokens": 284,
            "reasoning_output_tokens": 198
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34682,
            "output_tokens": 444,
            "reasoning_output_tokens": 327
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35500,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35727,
            "output_tokens": 80,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36204,
            "output_tokens": 339,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36572,
            "output_tokens": 538,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37133,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 41,
        "model_tool_calls_by_name": {
          "read": 17,
          "ls": 1,
          "bash": 1,
          "devices": 3,
          "exec": 12,
          "write": 3,
          "device_stop": 2,
          "device": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 711 rosbridge control frames; 60 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 7
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/open-microwave.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/workspace/.kinex/memos/open-microwave.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-24-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 77,
        "observed_images": 13
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1500,
        "max_control_steps": 1500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 60,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 286,
        "published_frames": 286,
        "decode_ok": true,
        "duration_s": 7.15,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "b9447ca9494fe162d7230ad7b77cbe7a1b2612c704ca6c821be1c660b62aed37",
        "source_sha256": {
          "third_person": "c4ce4377e4cc55733a41527e00b7c2cb3a3ccd337c0ba65f3de087595bbc998a",
          "left_wrist": "1b73055dab76f7ab9e582c45fe374cd41c447ffe497c45bdd0c06b5de0df272e",
          "right_wrist": "22354d1e36bbf47e7a317801a36b1bb0ff7b3ee174896be989cdcdf2d497f807"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "cead1d44cf69",
          "task": "robotwin/open-microwave-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 286,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 286,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28440000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 286,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 286,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28440000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 286,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 286,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28440000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-25-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/25",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "25",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 2
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.88,
        "step_limit": 400,
        "steps": 397,
        "success": false,
        "valid": true
      },
      "steps": 397,
      "simulation_time_s": 15.88,
      "wall_time_s": 1077.800645,
      "timing": {
        "simulation_time_s": 15.88,
        "wall_time_s": 1077.800645,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift the bottle with attached loop on cap and the plastic bottle with blue label simultaneously with separate arms.",
      "native_instruction": "Lift the bottle with attached loop on cap and the plastic bottle with blue label simultaneously with separate arms.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9693702103987698,
        "cache_read_input_tokens": 1516032,
        "cache_reported_input_tokens": 1563935,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1563935,
        "cached_input_tokens": 1516032,
        "cost_usd": null,
        "fresh_input_tokens": 47903,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1563935,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1516032,
        "known_fresh_input_tokens": 47903,
        "known_input_tokens": 1563935,
        "known_output_tokens": 9423,
        "known_reasoning_output_tokens": 2940,
        "output_tokens": 9423,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2940,
        "reasoning_reported_output_tokens": 9423,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 44,
          "cache_write_input_tokens": 44,
          "cache_write_reported_input_tokens": 44,
          "cached_input_tokens": 44,
          "fresh_input_tokens": 44,
          "input_tokens": 44,
          "output_tokens": 44,
          "reasoning_output_tokens": 44,
          "reasoning_reported_output_tokens": 44
        },
        "request_attempts": 44,
        "response_count": 44,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1573358,
        "uncached_input_tokens": 47903,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11828,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11868,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12001,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12051,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12344,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 12495,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 18239,
            "output_tokens": 64,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 21488,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 22058,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22386,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22576,
            "output_tokens": 175,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 25745,
            "output_tokens": 314,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 27967,
            "output_tokens": 196,
            "reasoning_output_tokens": 156
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28629,
            "output_tokens": 746,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29398,
            "output_tokens": 171,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 35632,
            "output_tokens": 230,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 35891,
            "output_tokens": 148,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 37723,
            "output_tokens": 65,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37998,
            "output_tokens": 707,
            "reasoning_output_tokens": 104
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38727,
            "output_tokens": 141,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39178,
            "output_tokens": 324,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 40066,
            "output_tokens": 66,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40194,
            "output_tokens": 180,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40663,
            "output_tokens": 273,
            "reasoning_output_tokens": 168
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41218,
            "output_tokens": 120,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41510,
            "output_tokens": 270,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42123,
            "output_tokens": 253,
            "reasoning_output_tokens": 214
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 43071,
            "output_tokens": 674,
            "reasoning_output_tokens": 324
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 43940,
            "output_tokens": 138,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44321,
            "output_tokens": 136,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44479,
            "output_tokens": 312,
            "reasoning_output_tokens": 215
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45078,
            "output_tokens": 329,
            "reasoning_output_tokens": 117
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45813,
            "output_tokens": 224,
            "reasoning_output_tokens": 124
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46283,
            "output_tokens": 178,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46815,
            "output_tokens": 151,
            "reasoning_output_tokens": 110
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 46988,
            "output_tokens": 254,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 47497,
            "output_tokens": 377,
            "reasoning_output_tokens": 175
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 48102,
            "output_tokens": 305,
            "reasoning_output_tokens": 212
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 48651,
            "output_tokens": 381,
            "reasoning_output_tokens": 175
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49370,
            "output_tokens": 186,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49408,
            "input_tokens": 49868,
            "output_tokens": 156,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49920,
            "input_tokens": 50046,
            "output_tokens": 107,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50048,
            "input_tokens": 50550,
            "output_tokens": 493,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50944,
            "input_tokens": 51067,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 52,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 17,
          "bash": 1,
          "exec": 22,
          "devices": 1,
          "device": 1,
          "write": 3,
          "edit": 1,
          "device_stop": 4
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 397 rosbridge control frames; 35 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 9
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/bottle_manipulation.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/workspace/.kinex/memos/bottle_manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_tcp_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/workspace/.kinex/tools/dual_tcp_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-25-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 99,
        "observed_images": 28
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 35,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 160,
        "published_frames": 160,
        "decode_ok": true,
        "duration_s": 4.0,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "2d8dd3cda1c8d69d3cdec6dbecceacf93e812ba6f807aafbc132b2a6f0e12e61",
        "source_sha256": {
          "third_person": "2962ecbc0205a003c59d6a4c807153b42303b222850d2e949839a82ce1877cb0",
          "left_wrist": "09b4feb6ae0e04f9d6222515c42a0c24b4b2228fc5bcb2f2458c5ab8502f7c2c",
          "right_wrist": "1f7f38d835e569f79d2dbbb361421908dd6f7a483f2dd9c8b34e5db309086a09"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "d8317d809566",
          "task": "robotwin/pick-diverse-bottles-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 160,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 160,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15880000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 160,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 160,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15880000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 160,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 160,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15880000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-26-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/26",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "26",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 9.6,
        "step_limit": 400,
        "steps": 240,
        "success": false,
        "valid": true
      },
      "steps": 240,
      "simulation_time_s": 9.6,
      "wall_time_s": 717.325355,
      "timing": {
        "simulation_time_s": 9.6,
        "wall_time_s": 717.325355,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use one arm for the bottle with red screw cap and the other for the green bottle.",
      "native_instruction": "Use one arm for the bottle with red screw cap and the other for the green bottle.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9014147436322574,
        "cache_read_input_tokens": 516480,
        "cache_reported_input_tokens": 572966,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 572966,
        "cached_input_tokens": 516480,
        "cost_usd": null,
        "fresh_input_tokens": 56486,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 572966,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 516480,
        "known_fresh_input_tokens": 56486,
        "known_input_tokens": 572966,
        "known_output_tokens": 5116,
        "known_reasoning_output_tokens": 739,
        "output_tokens": 5116,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 739,
        "reasoning_reported_output_tokens": 5116,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 23,
          "cache_write_input_tokens": 23,
          "cache_write_reported_input_tokens": 23,
          "cached_input_tokens": 23,
          "fresh_input_tokens": 23,
          "input_tokens": 23,
          "output_tokens": 23,
          "reasoning_output_tokens": 23,
          "reasoning_reported_output_tokens": 23
        },
        "request_attempts": 23,
        "response_count": 23,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 578082,
        "uncached_input_tokens": 56486,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11820,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11986,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12034,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12327,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18216,
            "output_tokens": 166,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 22033,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22361,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22552,
            "output_tokens": 184,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 25477,
            "output_tokens": 294,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26279,
            "output_tokens": 1021,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27323,
            "output_tokens": 111,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 28511,
            "output_tokens": 94,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28815,
            "output_tokens": 192,
            "reasoning_output_tokens": 152
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29473,
            "output_tokens": 516,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30184,
            "output_tokens": 66,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 31239,
            "output_tokens": 900,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32161,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32665,
            "output_tokens": 94,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 33129,
            "output_tokens": 172,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33672,
            "output_tokens": 115,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34157,
            "output_tokens": 516,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34697,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 34,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 17,
          "bash": 1,
          "exec": 9,
          "devices": 1,
          "device": 1,
          "write": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 240 rosbridge control frames; 30 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 4
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/dual_bottle_pick.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/workspace/.kinex/memos/dual_bottle_pick.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_tcp_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/workspace/.kinex/tools/dual_tcp_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-26-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 60,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 30,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 97,
        "published_frames": 97,
        "decode_ok": true,
        "duration_s": 2.425,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "1cbb4d9d113ae99bb8d9f1849e7d4febe495d75943c10dd3a8c39fc0532caa34",
        "source_sha256": {
          "third_person": "fea7bd8906dfe1a38c4abab66b4281b6b8861da95a222fb576daee44010874a1",
          "left_wrist": "59083a66c665276a8b96f2eef46f54b106808f68e309aaf9faeefc3cbe1e7969",
          "right_wrist": "b417ff24522e4d9397826e38886c2f572009f7cac1baa6acd5e25612960d9fed"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "2740912ac418",
          "task": "robotwin/pick-dual-bottles-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 97,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 97,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 97,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-27-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/27",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "27",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 11.36,
        "step_limit": 400,
        "steps": 284,
        "success": true,
        "valid": true
      },
      "steps": 284,
      "simulation_time_s": 11.36,
      "wall_time_s": 721.641296,
      "timing": {
        "simulation_time_s": 11.36,
        "wall_time_s": 721.641296,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Position the coffee-box with mug image on the left of the mouse with scroll wheel",
      "native_instruction": "Position the coffee-box with mug image on the left of the mouse with scroll wheel",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9494580409422793,
        "cache_read_input_tokens": 735360,
        "cache_reported_input_tokens": 774505,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 774505,
        "cached_input_tokens": 735360,
        "cost_usd": null,
        "fresh_input_tokens": 39145,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 774505,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 735360,
        "known_fresh_input_tokens": 39145,
        "known_input_tokens": 774505,
        "known_output_tokens": 5011,
        "known_reasoning_output_tokens": 598,
        "output_tokens": 5011,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 598,
        "reasoning_reported_output_tokens": 5011,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 30,
          "cache_write_input_tokens": 30,
          "cache_write_reported_input_tokens": 30,
          "cached_input_tokens": 30,
          "fresh_input_tokens": 30,
          "input_tokens": 30,
          "output_tokens": 30,
          "reasoning_output_tokens": 30,
          "reasoning_reported_output_tokens": 30
        },
        "request_attempts": 30,
        "response_count": 30,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 779516,
        "uncached_input_tokens": 39145,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11821,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11861,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11994,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 12042,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12335,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 16591,
            "output_tokens": 188,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 20422,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 22062,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22390,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 22580,
            "output_tokens": 64,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 23013,
            "output_tokens": 299,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 25257,
            "output_tokens": 295,
            "reasoning_output_tokens": 253
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26591,
            "output_tokens": 810,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27424,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28091,
            "output_tokens": 68,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28369,
            "output_tokens": 103,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 29651,
            "output_tokens": 51,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29764,
            "output_tokens": 714,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30499,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30891,
            "output_tokens": 107,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31318,
            "output_tokens": 185,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31801,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32195,
            "output_tokens": 164,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32680,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33074,
            "output_tokens": 229,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33445,
            "output_tokens": 56,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33563,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33756,
            "output_tokens": 271,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34231,
            "output_tokens": 538,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34794,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 38,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "bash": 1,
          "exec": 11,
          "devices": 1,
          "device": 2,
          "write": 3,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 284 rosbridge control frames; 22 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 7
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-27-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 71,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 22,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 115,
        "published_frames": 115,
        "decode_ok": true,
        "duration_s": 2.875,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "615053a1ed2eca8ccfffd16635ec918c79a0f301ee66e769d20cfda053c2bb7d",
        "source_sha256": {
          "third_person": "8b674018174b4771e0f5700de614af034a4cc75748f9f21aa6394cf7b51b9a8a",
          "left_wrist": "78166387e85feeb7feb61da7d002e0e67d1472770c013a1b842450c8c8590417",
          "right_wrist": "845320bada68690761788fde034679dc113bc3de9f5a96ceb7209680230794d4"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "1366666fc4b2",
          "task": "robotwin/place-a2b-left-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 115,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 115,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11360000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 115,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 115,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11360000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 115,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 115,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11360000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-28-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/28",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "28",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.96,
        "step_limit": 400,
        "steps": 399,
        "success": false,
        "valid": true
      },
      "steps": 399,
      "simulation_time_s": 15.96,
      "wall_time_s": 772.767144,
      "timing": {
        "simulation_time_s": 15.96,
        "wall_time_s": 772.767144,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Put the tiny pink car with four wheels to the right of the red, blue, green, yellow rubikscube",
      "native_instruction": "Put the tiny pink car with four wheels to the right of the red, blue, green, yellow rubikscube",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9705373692518996,
        "cache_read_input_tokens": 2252032,
        "cache_reported_input_tokens": 2320397,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2320397,
        "cached_input_tokens": 2252032,
        "cost_usd": null,
        "fresh_input_tokens": 68365,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2320397,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2252032,
        "known_fresh_input_tokens": 68365,
        "known_input_tokens": 2320397,
        "known_output_tokens": 7777,
        "known_reasoning_output_tokens": 2163,
        "output_tokens": 7777,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2163,
        "reasoning_reported_output_tokens": 7777,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 51,
          "cache_write_input_tokens": 51,
          "cache_write_reported_input_tokens": 51,
          "cached_input_tokens": 51,
          "fresh_input_tokens": 51,
          "input_tokens": 51,
          "output_tokens": 51,
          "reasoning_output_tokens": 51,
          "reasoning_reported_output_tokens": 51
        },
        "request_attempts": 51,
        "response_count": 51,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2328174,
        "uncached_input_tokens": 68365,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11830,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11963,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12011,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12304,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 12455,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 12534,
            "output_tokens": 166,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 16650,
            "output_tokens": 53,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 17234,
            "output_tokens": 178,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 20727,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 20957,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21622,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 21812,
            "output_tokens": 151,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 43532,
            "output_tokens": 57,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43652,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 44007,
            "output_tokens": 354,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44901,
            "output_tokens": 136,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 46216,
            "output_tokens": 74,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46756,
            "output_tokens": 932,
            "reasoning_output_tokens": 117
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47488,
            "input_tokens": 47709,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 47865,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 48498,
            "output_tokens": 132,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 48657,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48512,
            "input_tokens": 49450,
            "output_tokens": 54,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49280,
            "input_tokens": 50121,
            "output_tokens": 98,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50048,
            "input_tokens": 50246,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50176,
            "input_tokens": 51058,
            "output_tokens": 304,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51200,
            "input_tokens": 51389,
            "output_tokens": 101,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51328,
            "input_tokens": 52249,
            "output_tokens": 56,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52096,
            "input_tokens": 53093,
            "output_tokens": 201,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53120,
            "input_tokens": 53321,
            "output_tokens": 101,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53248,
            "input_tokens": 54178,
            "output_tokens": 134,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54144,
            "input_tokens": 54720,
            "output_tokens": 301,
            "reasoning_output_tokens": 200
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54912,
            "input_tokens": 55437,
            "output_tokens": 181,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55552,
            "input_tokens": 56038,
            "output_tokens": 80,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55936,
            "input_tokens": 56423,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56320,
            "input_tokens": 56926,
            "output_tokens": 193,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56704,
            "input_tokens": 57535,
            "output_tokens": 159,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57472,
            "input_tokens": 58002,
            "output_tokens": 138,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57984,
            "input_tokens": 59037,
            "output_tokens": 143,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59008,
            "input_tokens": 59482,
            "output_tokens": 362,
            "reasoning_output_tokens": 275
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59648,
            "input_tokens": 60147,
            "output_tokens": 425,
            "reasoning_output_tokens": 339
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60416,
            "input_tokens": 60640,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60544,
            "input_tokens": 61045,
            "output_tokens": 186,
            "reasoning_output_tokens": 96
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61056,
            "input_tokens": 61538,
            "output_tokens": 159,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61568,
            "input_tokens": 62004,
            "output_tokens": 168,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62080,
            "input_tokens": 62476,
            "output_tokens": 225,
            "reasoning_output_tokens": 141
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62592,
            "input_tokens": 63006,
            "output_tokens": 273,
            "reasoning_output_tokens": 138
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 63104,
            "input_tokens": 63697,
            "output_tokens": 80,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 63616,
            "input_tokens": 64195,
            "output_tokens": 45,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 64000,
            "input_tokens": 64302,
            "output_tokens": 426,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 64512,
            "input_tokens": 64750,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 58,
        "model_tool_calls_by_name": {
          "read": 16,
          "ls": 1,
          "bash": 2,
          "devices": 3,
          "exec": 27,
          "device": 1,
          "write": 2,
          "edit": 4,
          "grep": 1,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 399 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 21
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/workspace/.kinex/memos/robotwin.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-28-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 112,
        "observed_images": 27
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 161,
        "published_frames": 161,
        "decode_ok": true,
        "duration_s": 4.025,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "a48eea0657f88d528954fe8782d91bade71d8e39455c31e3eade7619a5a0c604",
        "source_sha256": {
          "third_person": "48061ac366555c728c3fb7d17f43b0f14e9d18fbfac14efca6d9f602ff681638",
          "left_wrist": "5727a8619c74cd584f041bf62a895126f85a6784e3942e3793df9bc663c18c77",
          "right_wrist": "df8d804ec11aa6f63510f37213628fd60af660aad469ecf2981d200df240e67f"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "15d7e335a0b4",
          "task": "robotwin/place-a2b-right-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": false
    },
    {
      "id": "kinex-pose-film-29-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/29",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "29",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 25.68,
        "step_limit": 700,
        "steps": 642,
        "success": false,
        "valid": true
      },
      "steps": 642,
      "simulation_time_s": 25.68,
      "wall_time_s": 1334.586469,
      "timing": {
        "simulation_time_s": 25.68,
        "wall_time_s": 1334.586469,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move both the soft bread with golden edges and the soft rectangular bread loaf inside the white basket with beige bottom.",
      "native_instruction": "Move both the soft bread with golden edges and the soft rectangular bread loaf inside the white basket with beige bottom.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9739206611223656,
        "cache_read_input_tokens": 1672064,
        "cache_reported_input_tokens": 1716838,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1716838,
        "cached_input_tokens": 1672064,
        "cost_usd": null,
        "fresh_input_tokens": 44774,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1716838,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1672064,
        "known_fresh_input_tokens": 44774,
        "known_input_tokens": 1716838,
        "known_output_tokens": 9294,
        "known_reasoning_output_tokens": 2807,
        "output_tokens": 9294,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2807,
        "reasoning_reported_output_tokens": 9294,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 51,
          "cache_write_input_tokens": 51,
          "cache_write_reported_input_tokens": 51,
          "cached_input_tokens": 51,
          "fresh_input_tokens": 51,
          "input_tokens": 51,
          "output_tokens": 51,
          "reasoning_output_tokens": 51,
          "reasoning_reported_output_tokens": 51
        },
        "request_attempts": 51,
        "response_count": 51,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1726132,
        "uncached_input_tokens": 44774,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11827,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11867,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12000,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12344,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 12807,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12672,
            "input_tokens": 13621,
            "output_tokens": 195,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 18245,
            "output_tokens": 134,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 20448,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 22088,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22416,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22605,
            "output_tokens": 65,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 23709,
            "output_tokens": 316,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24420,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24828,
            "output_tokens": 311,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 26880,
            "output_tokens": 1045,
            "reasoning_output_tokens": 168
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 27947,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28714,
            "output_tokens": 151,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29529,
            "output_tokens": 64,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29655,
            "output_tokens": 61,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29926,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30955,
            "output_tokens": 152,
            "reasoning_output_tokens": 108
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 32930,
            "output_tokens": 953,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 33904,
            "output_tokens": 119,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34541,
            "output_tokens": 234,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34996,
            "output_tokens": 190,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35707,
            "output_tokens": 106,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36332,
            "output_tokens": 79,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36473,
            "output_tokens": 98,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 37074,
            "output_tokens": 106,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37705,
            "output_tokens": 483,
            "reasoning_output_tokens": 369
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38706,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39294,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39844,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40433,
            "output_tokens": 81,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40576,
            "output_tokens": 75,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40976,
            "output_tokens": 336,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41590,
            "output_tokens": 175,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42280,
            "output_tokens": 161,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 43339,
            "output_tokens": 317,
            "reasoning_output_tokens": 234
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 44167,
            "output_tokens": 92,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44762,
            "output_tokens": 100,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45386,
            "output_tokens": 301,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 46015,
            "output_tokens": 193,
            "reasoning_output_tokens": 117
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46720,
            "output_tokens": 320,
            "reasoning_output_tokens": 243
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47550,
            "output_tokens": 299,
            "reasoning_output_tokens": 224
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 48358,
            "output_tokens": 88,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 48956,
            "output_tokens": 98,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49566,
            "output_tokens": 154,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49664,
            "input_tokens": 49888,
            "output_tokens": 214,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50048,
            "input_tokens": 50617,
            "output_tokens": 681,
            "reasoning_output_tokens": 177
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51200,
            "input_tokens": 51322,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 57,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 16,
          "bash": 1,
          "exec": 28,
          "devices": 2,
          "device": 2,
          "write": 3,
          "device_stop": 3,
          "grep": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 642 rosbridge control frames; 41 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 19
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/place_bread_basket.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/workspace/.kinex/memos/place_bread_basket.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-29-20261116/workspace/.kinex/tools/tcp_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 111,
        "observed_images": 30
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 41,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 258,
        "published_frames": 258,
        "decode_ok": true,
        "duration_s": 6.45,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "30bcb2f850b31eafd416afbeff4e5c8f8d72dffa732a74729e10badffbadbcb7",
        "source_sha256": {
          "third_person": "eb382aba204beca04ad606f81059a1af427273fa425e7951717ce9584941a775",
          "left_wrist": "15d540d80aa0a0d45187eb68eebf739adb3e23ba4d16d8fe236f29f4a3c98096",
          "right_wrist": "eaebd0448cc8db2b3fc52c5c610199504daaa750ec8465b816944d316e505a97"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "1de645e9e5ac",
          "task": "robotwin/place-bread-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 258,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 258,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 25680000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 258,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 258,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 25680000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 258,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 258,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 25680000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-30-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/30",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "30",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 14.200000000000001,
        "step_limit": 500,
        "steps": 355,
        "success": true,
        "valid": true
      },
      "steps": 355,
      "simulation_time_s": 14.200000000000001,
      "wall_time_s": 922.417835,
      "timing": {
        "simulation_time_s": 14.200000000000001,
        "wall_time_s": 922.417835,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move the spongy golden brown bread from the table into the versatile skillet for frying and sauteing.",
      "native_instruction": "Move the spongy golden brown bread from the table into the versatile skillet for frying and sauteing.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9109774545278521,
        "cache_read_input_tokens": 752768,
        "cache_reported_input_tokens": 826330,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 826330,
        "cached_input_tokens": 752768,
        "cost_usd": null,
        "fresh_input_tokens": 73562,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 826330,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 752768,
        "known_fresh_input_tokens": 73562,
        "known_input_tokens": 826330,
        "known_output_tokens": 6178,
        "known_reasoning_output_tokens": 1426,
        "output_tokens": 6178,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1426,
        "reasoning_reported_output_tokens": 6178,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 30,
          "cache_write_input_tokens": 30,
          "cache_write_reported_input_tokens": 30,
          "cached_input_tokens": 30,
          "fresh_input_tokens": 30,
          "input_tokens": 30,
          "output_tokens": 30,
          "reasoning_output_tokens": 30,
          "reasoning_reported_output_tokens": 30
        },
        "request_attempts": 30,
        "response_count": 30,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 832508,
        "uncached_input_tokens": 73562,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11822,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11955,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12003,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12296,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18185,
            "output_tokens": 240,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 20711,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 22111,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22300,
            "output_tokens": 67,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22736,
            "output_tokens": 273,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23529,
            "output_tokens": 1041,
            "reasoning_output_tokens": 193
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24593,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 26024,
            "output_tokens": 109,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26343,
            "output_tokens": 143,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 27813,
            "output_tokens": 45,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 27920,
            "output_tokens": 88,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28997,
            "output_tokens": 869,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29887,
            "output_tokens": 143,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30956,
            "output_tokens": 306,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 31483,
            "output_tokens": 226,
            "reasoning_output_tokens": 186
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 32175,
            "output_tokens": 123,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 33220,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 34002,
            "output_tokens": 118,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34596,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 35375,
            "output_tokens": 127,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36197,
            "output_tokens": 638,
            "reasoning_output_tokens": 294
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37240,
            "output_tokens": 243,
            "reasoning_output_tokens": 140
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 37628,
            "output_tokens": 55,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37745,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 37937,
            "output_tokens": 589,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38551,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 39,
        "model_tool_calls_by_name": {
          "read": 17,
          "ls": 1,
          "bash": 1,
          "devices": 1,
          "exec": 12,
          "device": 2,
          "write": 3,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 355 rosbridge control frames; 40 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 6
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-30-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 72,
        "observed_images": 13
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 40,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 143,
        "published_frames": 143,
        "decode_ok": true,
        "duration_s": 3.575,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "4438712933e05ab56eca645b7947af363f8fa46325d81f24cf6786fdf3bed7f0",
        "source_sha256": {
          "third_person": "c6a4eeaaf041e3e15b03e54667bf60ba27fb06333aac4ee5346ff48e89972105",
          "left_wrist": "487a6f3c412d6ed6553e5ff89057c6adf1719344474660eeee02047e8025e971",
          "right_wrist": "45d74b0d699eee5426515e048ac46511acaa1f6a582fdedf8ec48108cf945570"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "eb7235436b35",
          "task": "robotwin/place-bread-skillet-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 144,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 143,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 144,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 143,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 144,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 143,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-31-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/31",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "31",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 16.04,
        "step_limit": 500,
        "steps": 401,
        "success": true,
        "valid": true
      },
      "steps": 401,
      "simulation_time_s": 16.04,
      "wall_time_s": 1066.725273,
      "timing": {
        "simulation_time_s": 16.04,
        "wall_time_s": 1066.725273,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the compact hamburg box and the Crispy potato fries with red box, place them on the green tray.",
      "native_instruction": "Pick the compact hamburg box and the Crispy potato fries with red box, place them on the green tray.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9682439037905144,
        "cache_read_input_tokens": 1223168,
        "cache_reported_input_tokens": 1263285,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1263285,
        "cached_input_tokens": 1223168,
        "cost_usd": null,
        "fresh_input_tokens": 40117,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1263285,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1223168,
        "known_fresh_input_tokens": 40117,
        "known_input_tokens": 1263285,
        "known_output_tokens": 7415,
        "known_reasoning_output_tokens": 1311,
        "output_tokens": 7415,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1311,
        "reasoning_reported_output_tokens": 7415,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 48,
          "cache_write_input_tokens": 48,
          "cache_write_reported_input_tokens": 48,
          "cached_input_tokens": 48,
          "fresh_input_tokens": 48,
          "input_tokens": 48,
          "output_tokens": 48,
          "reasoning_output_tokens": 48,
          "reasoning_reported_output_tokens": 48
        },
        "request_attempts": 48,
        "response_count": 48,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1270700,
        "uncached_input_tokens": 40117,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11822,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11862,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11995,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12045,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12338,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 12801,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12672,
            "input_tokens": 13615,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13568,
            "input_tokens": 14432,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14336,
            "input_tokens": 15258,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15104,
            "input_tokens": 16086,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 16878,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16768,
            "input_tokens": 18102,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18253,
            "output_tokens": 57,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 18841,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 19646,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20450,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20778,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 22178,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22586,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 23135,
            "output_tokens": 321,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23906,
            "output_tokens": 291,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 25011,
            "output_tokens": 260,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25593,
            "output_tokens": 83,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 26142,
            "output_tokens": 198,
            "reasoning_output_tokens": 156
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 27379,
            "output_tokens": 869,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28271,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 29014,
            "output_tokens": 162,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 30011,
            "output_tokens": 46,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30119,
            "output_tokens": 109,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30438,
            "output_tokens": 849,
            "reasoning_output_tokens": 217
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31308,
            "output_tokens": 124,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 32022,
            "output_tokens": 289,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32635,
            "output_tokens": 180,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33193,
            "output_tokens": 113,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33902,
            "output_tokens": 93,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34367,
            "output_tokens": 182,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 35146,
            "output_tokens": 321,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35755,
            "output_tokens": 176,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36527,
            "output_tokens": 52,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36641,
            "output_tokens": 101,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37338,
            "output_tokens": 114,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 38044,
            "output_tokens": 246,
            "reasoning_output_tokens": 108
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38664,
            "output_tokens": 170,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39207,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39407,
            "output_tokens": 54,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39523,
            "output_tokens": 152,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39899,
            "output_tokens": 799,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40722,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 47,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "bash": 1,
          "devices": 1,
          "exec": 21,
          "write": 3,
          "device_stop": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 401 rosbridge control frames; 40 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 11
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/aloha_tabletop.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/workspace/.kinex/memos/aloha_tabletop.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-31-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 98,
        "observed_images": 30
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 40,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 162,
        "published_frames": 162,
        "decode_ok": true,
        "duration_s": 4.05,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "a2df4b30d0be4d4c09bfc44ed7850d825d91600f2a40c5c94051e000e88eb13d",
        "source_sha256": {
          "third_person": "7896cebc2df06fce688cc20d27a0f2bc2ae46e221938e7a49bca15efffa056de",
          "left_wrist": "8b29196e3d392316d2240c73cf98cbdeecd20143fba14a33f5d3eb948aa057fa",
          "right_wrist": "3760557150cdb805b3feda8e08d06dea19b6468b25b06004b5a51ac2a03ab653"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "6dbc2de05dea",
          "task": "robotwin/place-burger-fries-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 162,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 162,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16040000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 162,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 162,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16040000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 162,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 162,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 16040000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-32-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/32",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "32",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 19.52,
        "step_limit": 700,
        "steps": 488,
        "success": true,
        "valid": true
      },
      "steps": 488,
      "simulation_time_s": 19.52,
      "wall_time_s": 1211.896723,
      "timing": {
        "simulation_time_s": 19.52,
        "wall_time_s": 1211.896723,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to put the red can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "native_instruction": "Use the left arm to put the red can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9666029461374115,
        "cache_read_input_tokens": 1375360,
        "cache_reported_input_tokens": 1422880,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1422880,
        "cached_input_tokens": 1375360,
        "cost_usd": null,
        "fresh_input_tokens": 47520,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1422880,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1375360,
        "known_fresh_input_tokens": 47520,
        "known_input_tokens": 1422880,
        "known_output_tokens": 9007,
        "known_reasoning_output_tokens": 2661,
        "output_tokens": 9007,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2661,
        "reasoning_reported_output_tokens": 9007,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 41,
          "cache_write_input_tokens": 41,
          "cache_write_reported_input_tokens": 41,
          "cached_input_tokens": 41,
          "fresh_input_tokens": 41,
          "input_tokens": 41,
          "output_tokens": 41,
          "reasoning_output_tokens": 41,
          "reasoning_reported_output_tokens": 41
        },
        "request_attempts": 41,
        "response_count": 41,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1431887,
        "uncached_input_tokens": 47520,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11828,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11961,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12009,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12302,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 18191,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 19831,
            "output_tokens": 220,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 22337,
            "output_tokens": 169,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 25412,
            "output_tokens": 297,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26442,
            "output_tokens": 310,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27886,
            "output_tokens": 81,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28433,
            "output_tokens": 1108,
            "reasoning_output_tokens": 226
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29564,
            "output_tokens": 142,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30651,
            "output_tokens": 76,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30774,
            "output_tokens": 145,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31129,
            "output_tokens": 252,
            "reasoning_output_tokens": 110
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 32328,
            "output_tokens": 71,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 33578,
            "output_tokens": 874,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34473,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 35275,
            "output_tokens": 78,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 37185,
            "output_tokens": 162,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37374,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 38347,
            "output_tokens": 187,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38905,
            "output_tokens": 125,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39401,
            "output_tokens": 178,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 40445,
            "output_tokens": 132,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40948,
            "output_tokens": 74,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41084,
            "output_tokens": 92,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41500,
            "output_tokens": 251,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41977,
            "output_tokens": 404,
            "reasoning_output_tokens": 270
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42751,
            "output_tokens": 110,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43234,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43677,
            "output_tokens": 127,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44172,
            "output_tokens": 445,
            "reasoning_output_tokens": 200
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 45109,
            "output_tokens": 487,
            "reasoning_output_tokens": 354
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45971,
            "output_tokens": 109,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46456,
            "output_tokens": 653,
            "reasoning_output_tokens": 452
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47371,
            "output_tokens": 148,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47488,
            "input_tokens": 47661,
            "output_tokens": 71,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 47794,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 48223,
            "output_tokens": 645,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48768,
            "input_tokens": 48891,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 53,
        "model_tool_calls_by_name": {
          "read": 18,
          "ls": 1,
          "bash": 1,
          "exec": 22,
          "devices": 3,
          "device": 1,
          "write": 3,
          "grep": 1,
          "edit": 1,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 488 rosbridge control frames; 42 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 13
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/place-can-basket.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/workspace/.kinex/memos/place-can-basket.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-32-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 97,
        "observed_images": 23
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 42,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 196,
        "published_frames": 196,
        "decode_ok": true,
        "duration_s": 4.9,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "6f6196c0a05584dbdc9c678e6447076f220f883d00019fa460a00c32c9b2784b",
        "source_sha256": {
          "third_person": "a045c67779ebbed70824057a9c275fae677cd95657e6870de7db529108baaec2",
          "left_wrist": "5d2a7715683a33ddf9f6f2a3d7495cc7cc227e6e0b22ea5ddabf9b93321a3273",
          "right_wrist": "aa07349a7fbb9dcc535da0e717ff54c11861f0d1bbb5545537e4fcd7be4d5a06"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "01ffbbe24a1e",
          "task": "robotwin/place-can-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 197,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 196,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19520000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 197,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 196,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19520000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 197,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 196,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19520000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-33-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/33",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "33",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 24.16,
        "step_limit": 800,
        "steps": 604,
        "success": false,
        "valid": true
      },
      "steps": 604,
      "simulation_time_s": 24.16,
      "wall_time_s": 1318.659532,
      "timing": {
        "simulation_time_s": 24.16,
        "wall_time_s": 1318.659532,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the small red soda can and the lightweight cylindrical metal can in parallel, then put both into the rectangular plasticbox.",
      "native_instruction": "Grab the small red soda can and the lightweight cylindrical metal can in parallel, then put both into the rectangular plasticbox.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9743750416555844,
        "cache_read_input_tokens": 1564288,
        "cache_reported_input_tokens": 1605427,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1605427,
        "cached_input_tokens": 1564288,
        "cost_usd": null,
        "fresh_input_tokens": 41139,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1605427,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1564288,
        "known_fresh_input_tokens": 41139,
        "known_input_tokens": 1605427,
        "known_output_tokens": 9338,
        "known_reasoning_output_tokens": 2205,
        "output_tokens": 9338,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2205,
        "reasoning_reported_output_tokens": 9338,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 49,
          "cache_write_input_tokens": 49,
          "cache_write_reported_input_tokens": 49,
          "cached_input_tokens": 49,
          "fresh_input_tokens": 49,
          "input_tokens": 49,
          "output_tokens": 49,
          "reasoning_output_tokens": 49,
          "reasoning_reported_output_tokens": 49
        },
        "request_attempts": 49,
        "response_count": 49,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1614765,
        "uncached_input_tokens": 41139,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11825,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11865,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11998,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 12048,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 14963,
            "output_tokens": 169,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14976,
            "input_tokens": 16906,
            "output_tokens": 183,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 20761,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21950,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22278,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22686,
            "output_tokens": 204,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23081,
            "output_tokens": 172,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 25667,
            "output_tokens": 305,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26295,
            "output_tokens": 1165,
            "reasoning_output_tokens": 280
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27484,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28210,
            "output_tokens": 197,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 29019,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29112,
            "output_tokens": 114,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 30423,
            "output_tokens": 963,
            "reasoning_output_tokens": 92
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31410,
            "output_tokens": 125,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31915,
            "output_tokens": 292,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32528,
            "output_tokens": 263,
            "reasoning_output_tokens": 190
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32859,
            "output_tokens": 122,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33370,
            "output_tokens": 73,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33505,
            "output_tokens": 82,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 34053,
            "output_tokens": 129,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34469,
            "output_tokens": 256,
            "reasoning_output_tokens": 131
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35103,
            "output_tokens": 145,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 36231,
            "output_tokens": 190,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36799,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37055,
            "output_tokens": 418,
            "reasoning_output_tokens": 171
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37725,
            "output_tokens": 285,
            "reasoning_output_tokens": 198
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38389,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38645,
            "output_tokens": 324,
            "reasoning_output_tokens": 271
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39928,
            "output_tokens": 238,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40537,
            "output_tokens": 77,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40636,
            "output_tokens": 102,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 41118,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41574,
            "output_tokens": 268,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42004,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42769,
            "output_tokens": 96,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43250,
            "output_tokens": 104,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43733,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44161,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44928,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45386,
            "output_tokens": 258,
            "reasoning_output_tokens": 220
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45706,
            "output_tokens": 278,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46014,
            "output_tokens": 204,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46248,
            "output_tokens": 534,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 46808,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 58,
        "model_tool_calls_by_name": {
          "ls": 5,
          "read": 13,
          "find": 1,
          "exec": 28,
          "devices": 1,
          "write": 3,
          "device_stop": 4,
          "grep": 1,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 604 rosbridge control frames; 59 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 14
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-cans.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/workspace/.kinex/memos/robotwin-cans.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_cartesian.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/workspace/.kinex/tools/dual_cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-33-20261116/workspace/.kinex/tools/dual_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 110,
        "observed_images": 25
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 59,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 243,
        "published_frames": 243,
        "decode_ok": true,
        "duration_s": 6.075,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "2ff8fb3175f4a43d02939e031abdb3de13c4298436f78f88fcd13363d4427d84",
        "source_sha256": {
          "third_person": "016aac4486318c6185b5edeef6bc65c8a3bca3ec7ef06ae86160e6ae796edb69",
          "left_wrist": "d9bc691b3593fc1325c3177fe27cbbe4413e074d5f3b078f382549b38938325c",
          "right_wrist": "b209a9328a5314d4cb9b8965be134664a4e1db1a9f04abc246e3719e159f6199"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "390f00904440",
          "task": "robotwin/place-cans-plasticbox-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 243,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 243,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24160000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 243,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 243,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24160000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 243,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 243,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 24160000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-34-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/34",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "34",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.200000000000001,
        "step_limit": 400,
        "steps": 305,
        "success": true,
        "valid": true
      },
      "steps": 305,
      "simulation_time_s": 12.200000000000001,
      "wall_time_s": 817.147414,
      "timing": {
        "simulation_time_s": 12.200000000000001,
        "wall_time_s": 817.147414,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the ceramic drinking cup with floral design on the gray plate for dining.",
      "native_instruction": "Place the ceramic drinking cup with floral design on the gray plate for dining.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9355393695023551,
        "cache_read_input_tokens": 1003264,
        "cache_reported_input_tokens": 1072391,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1072391,
        "cached_input_tokens": 1003264,
        "cost_usd": null,
        "fresh_input_tokens": 69127,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1072391,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1003264,
        "known_fresh_input_tokens": 69127,
        "known_input_tokens": 1072391,
        "known_output_tokens": 6276,
        "known_reasoning_output_tokens": 1130,
        "output_tokens": 6276,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1130,
        "reasoning_reported_output_tokens": 6276,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 34,
          "cache_write_input_tokens": 34,
          "cache_write_reported_input_tokens": 34,
          "cached_input_tokens": 34,
          "fresh_input_tokens": 34,
          "input_tokens": 34,
          "output_tokens": 34,
          "reasoning_output_tokens": 34,
          "reasoning_reported_output_tokens": 34
        },
        "request_attempts": 34,
        "response_count": 34,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1078667,
        "uncached_input_tokens": 69127,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11859,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11992,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15588,
            "output_tokens": 201,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15616,
            "input_tokens": 18341,
            "output_tokens": 245,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 23775,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24103,
            "output_tokens": 156,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24655,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25063,
            "output_tokens": 158,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26952,
            "output_tokens": 333,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27759,
            "output_tokens": 1033,
            "reasoning_output_tokens": 153
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28816,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 30807,
            "output_tokens": 188,
            "reasoning_output_tokens": 104
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 32664,
            "output_tokens": 70,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32944,
            "output_tokens": 334,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 34146,
            "output_tokens": 130,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 34742,
            "output_tokens": 897,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35662,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36376,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 36481,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36884,
            "output_tokens": 66,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37012,
            "output_tokens": 66,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37252,
            "output_tokens": 137,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37706,
            "output_tokens": 75,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38130,
            "output_tokens": 101,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 38546,
            "output_tokens": 261,
            "reasoning_output_tokens": 149
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 39123,
            "output_tokens": 146,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39584,
            "output_tokens": 262,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 39875,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40276,
            "output_tokens": 67,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40489,
            "output_tokens": 52,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40603,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40872,
            "output_tokens": 599,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41495,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 45,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 15,
          "find": 1,
          "devices": 2,
          "exec": 15,
          "write": 3,
          "edit": 2,
          "device_stop": 2,
          "device": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 305 rosbridge control frames; 36 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 9
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/manipulation.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/workspace/.kinex/memos/manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-34-20261116/workspace/.kinex/tools/tcp_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 82,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 36,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 123,
        "published_frames": 123,
        "decode_ok": true,
        "duration_s": 3.075,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c3f7ee6864b1adc5d93b8cc04f8ce709a11435cca3f9eb734f4756dd1a5f49b1",
        "source_sha256": {
          "third_person": "f79c3194d76cc1f20d8a17f96d559cee6e74de2329cff788a59576201c51bd6b",
          "left_wrist": "9ad23e6e4287aa1b43792e516bf05d730525bb578a73f7a6d4053cdd1432627d",
          "right_wrist": "5e974bcaab587eb57cce62f43a0d82fd27a6b877fde91d3edf56320886ab9e10"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "c99b48cad4a8",
          "task": "robotwin/place-container-plate-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 124,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 123,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 124,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 123,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 124,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 123,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-35-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/35",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "35",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 19.72,
        "step_limit": 500,
        "steps": 493,
        "success": false,
        "valid": true
      },
      "steps": 493,
      "simulation_time_s": 19.72,
      "wall_time_s": 1418.671359,
      "timing": {
        "simulation_time_s": 19.72,
        "wall_time_s": 1418.671359,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the small smooth cup onto the light gray coaster gently.",
      "native_instruction": "Place the small smooth cup onto the light gray coaster gently.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9809266475617227,
        "cache_read_input_tokens": 2659968,
        "cache_reported_input_tokens": 2711689,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2711689,
        "cached_input_tokens": 2659968,
        "cost_usd": null,
        "fresh_input_tokens": 51721,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2711689,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2659968,
        "known_fresh_input_tokens": 51721,
        "known_input_tokens": 2711689,
        "known_output_tokens": 13229,
        "known_reasoning_output_tokens": 3907,
        "output_tokens": 13229,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3907,
        "reasoning_reported_output_tokens": 13229,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 76,
          "cache_write_input_tokens": 76,
          "cache_write_reported_input_tokens": 76,
          "cached_input_tokens": 76,
          "fresh_input_tokens": 76,
          "input_tokens": 76,
          "output_tokens": 76,
          "reasoning_output_tokens": 76,
          "reasoning_reported_output_tokens": 76
        },
        "request_attempts": 76,
        "response_count": 76,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2724918,
        "uncached_input_tokens": 51721,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11815,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11988,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12332,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12795,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12672,
            "input_tokens": 13609,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13440,
            "input_tokens": 14426,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14336,
            "input_tokens": 15254,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15104,
            "input_tokens": 16080,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 16872,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16768,
            "input_tokens": 18096,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18247,
            "output_tokens": 195,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 21846,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22174,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22365,
            "output_tokens": 65,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22799,
            "output_tokens": 256,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23709,
            "output_tokens": 313,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24263,
            "output_tokens": 384,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25347,
            "output_tokens": 104,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25917,
            "output_tokens": 964,
            "reasoning_output_tokens": 92
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26903,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27804,
            "output_tokens": 65,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28079,
            "output_tokens": 335,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28780,
            "output_tokens": 176,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 30135,
            "output_tokens": 931,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31087,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31664,
            "output_tokens": 133,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31824,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32630,
            "output_tokens": 84,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33409,
            "output_tokens": 225,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 34350,
            "output_tokens": 102,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 35167,
            "output_tokens": 334,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 36044,
            "output_tokens": 307,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36837,
            "output_tokens": 292,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37156,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37610,
            "output_tokens": 50,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37709,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37845,
            "output_tokens": 55,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 37922,
            "output_tokens": 409,
            "reasoning_output_tokens": 219
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38557,
            "output_tokens": 255,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39068,
            "output_tokens": 119,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39214,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39695,
            "output_tokens": 114,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40174,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40267,
            "output_tokens": 227,
            "reasoning_output_tokens": 165
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 41038,
            "output_tokens": 169,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41570,
            "output_tokens": 110,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41848,
            "output_tokens": 417,
            "reasoning_output_tokens": 155
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42474,
            "output_tokens": 226,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 43416,
            "output_tokens": 126,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43907,
            "output_tokens": 315,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44424,
            "output_tokens": 442,
            "reasoning_output_tokens": 349
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45227,
            "output_tokens": 141,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45390,
            "output_tokens": 67,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45627,
            "output_tokens": 367,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45952,
            "input_tokens": 46286,
            "output_tokens": 176,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46828,
            "output_tokens": 43,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 46889,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 46971,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47241,
            "output_tokens": 132,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47742,
            "output_tokens": 581,
            "reasoning_output_tokens": 289
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 48526,
            "output_tokens": 113,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48512,
            "input_tokens": 49008,
            "output_tokens": 55,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49081,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49163,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49432,
            "output_tokens": 215,
            "reasoning_output_tokens": 130
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 50017,
            "output_tokens": 530,
            "reasoning_output_tokens": 438
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50432,
            "input_tokens": 50912,
            "output_tokens": 45,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50816,
            "input_tokens": 50975,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50816,
            "input_tokens": 51057,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50944,
            "input_tokens": 51344,
            "output_tokens": 323,
            "reasoning_output_tokens": 160
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51456,
            "input_tokens": 51815,
            "output_tokens": 103,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51712,
            "input_tokens": 52286,
            "output_tokens": 262,
            "reasoning_output_tokens": 141
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52480,
            "input_tokens": 52911,
            "output_tokens": 55,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52864,
            "input_tokens": 53028,
            "output_tokens": 486,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53376,
            "input_tokens": 53537,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 78,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 17,
          "bash": 1,
          "exec": 33,
          "devices": 5,
          "device": 10,
          "write": 3,
          "edit": 3,
          "relative_pose": 1,
          "device_stop": 4
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 493 rosbridge control frames; 28 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 19
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/place_empty_cup.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/workspace/.kinex/memos/place_empty_cup.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-35-20261116/workspace/.kinex/tools/tcp_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 157,
        "observed_images": 21
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 28,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 198,
        "published_frames": 198,
        "decode_ok": true,
        "duration_s": 4.95,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "077fc6f8d368387eb4cff8ab5f32107d44f58a113cb4b590a00cd60daee7effa",
        "source_sha256": {
          "third_person": "a3bcb80bc21dba9d2e81b191034858b08b2f6f1b4f1975817bb98897e0065ee0",
          "left_wrist": "a74512f84b855335736f0b84871859f67c32d3269214a9636663ccd2eb60d73c",
          "right_wrist": "7e7d349cc222fd066d84ba47fe21256f24992c7c41db0eea7386fa0f977dcf02"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "9dfb0c3eb9de",
          "task": "robotwin/place-empty-cup-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 199,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 198,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19720000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 199,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 198,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19720000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 199,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 198,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19720000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-36-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/36",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "36",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.96,
        "step_limit": 400,
        "steps": 324,
        "success": true,
        "valid": true
      },
      "steps": 324,
      "simulation_time_s": 12.96,
      "wall_time_s": 897.018696,
      "timing": {
        "simulation_time_s": 12.96,
        "wall_time_s": 897.018696,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the beige circular blade fan, place it on the Pink mat, face it toward the robot",
      "native_instruction": "Pick the beige circular blade fan, place it on the Pink mat, face it toward the robot",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9584108663791033,
        "cache_read_input_tokens": 1057408,
        "cache_reported_input_tokens": 1103293,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1103293,
        "cached_input_tokens": 1057408,
        "cost_usd": null,
        "fresh_input_tokens": 45885,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1103293,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1057408,
        "known_fresh_input_tokens": 45885,
        "known_input_tokens": 1103293,
        "known_output_tokens": 6610,
        "known_reasoning_output_tokens": 1222,
        "output_tokens": 6610,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1222,
        "reasoning_reported_output_tokens": 6610,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1109903,
        "uncached_input_tokens": 45885,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11822,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11862,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11995,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15619,
            "output_tokens": 218,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 19178,
            "output_tokens": 202,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 23780,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 25180,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25508,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25699,
            "output_tokens": 66,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26134,
            "output_tokens": 249,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26944,
            "output_tokens": 78,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27342,
            "output_tokens": 313,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28105,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28980,
            "output_tokens": 395,
            "reasoning_output_tokens": 110
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29629,
            "output_tokens": 942,
            "reasoning_output_tokens": 154
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30594,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 34611,
            "output_tokens": 294,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35119,
            "output_tokens": 140,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 39207,
            "output_tokens": 159,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 41027,
            "output_tokens": 782,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41830,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42265,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42700,
            "output_tokens": 205,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43365,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43806,
            "output_tokens": 312,
            "reasoning_output_tokens": 195
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44214,
            "output_tokens": 106,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44658,
            "output_tokens": 436,
            "reasoning_output_tokens": 161
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 46140,
            "output_tokens": 154,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46639,
            "output_tokens": 229,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 47010,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 47102,
            "output_tokens": 43,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47315,
            "output_tokens": 577,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 47914,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 46,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 17,
          "find": 1,
          "exec": 16,
          "devices": 2,
          "device": 2,
          "write": 3,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 324 rosbridge control frames; 36 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 7
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/place-fan.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/workspace/.kinex/memos/place-fan.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-36-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 82,
        "observed_images": 19
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 36,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 131,
        "published_frames": 131,
        "decode_ok": true,
        "duration_s": 3.275,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "8eb804cf004b6225f017a19c9eff9dc1b92f365866ae006bbc573e035a4dbef9",
        "source_sha256": {
          "third_person": "c524eeed802e7c7fdc698bb5e8096138bbdb57000f7f78780a15c3316c179a95",
          "left_wrist": "229dd0e67cc80e8a79ded31350c497f42f6cecc6a5c284b1ef9b7cba2a213951",
          "right_wrist": "c1b3bb42f805fb23d5cfbcbd09af3078024e2ee662f40c1770630721fa005e64"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ffcfcc8f0c0c",
          "task": "robotwin/place-fan-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 131,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 131,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 131,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 131,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 131,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 131,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-37-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/37",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "37",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 2
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 10.08,
        "step_limit": 400,
        "steps": 252,
        "success": true,
        "valid": true
      },
      "steps": 252,
      "simulation_time_s": 10.08,
      "wall_time_s": 428.594823,
      "timing": {
        "simulation_time_s": 10.08,
        "wall_time_s": 428.594823,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the dark gray computer mouse with logo with the left arm and drop it onto the Cyan mat.",
      "native_instruction": "Pick up the dark gray computer mouse with logo with the left arm and drop it onto the Cyan mat.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9632619262522462,
        "cache_read_input_tokens": 581632,
        "cache_reported_input_tokens": 603815,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 603815,
        "cached_input_tokens": 581632,
        "cost_usd": null,
        "fresh_input_tokens": 22183,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 603815,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 581632,
        "known_fresh_input_tokens": 22183,
        "known_input_tokens": 603815,
        "known_output_tokens": 4287,
        "known_reasoning_output_tokens": 1055,
        "output_tokens": 4287,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1055,
        "reasoning_reported_output_tokens": 4287,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 25,
          "cache_write_input_tokens": 25,
          "cache_write_reported_input_tokens": 25,
          "cached_input_tokens": 25,
          "fresh_input_tokens": 25,
          "input_tokens": 25,
          "output_tokens": 25,
          "reasoning_output_tokens": 25,
          "reasoning_reported_output_tokens": 25
        },
        "request_attempts": 25,
        "response_count": 25,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 608102,
        "uncached_input_tokens": 22183,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11824,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11864,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11997,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15469,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15360,
            "input_tokens": 15762,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 20018,
            "output_tokens": 83,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 21309,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21879,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22043,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22234,
            "output_tokens": 150,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23425,
            "output_tokens": 315,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24795,
            "output_tokens": 295,
            "reasoning_output_tokens": 216
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 26079,
            "output_tokens": 848,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26949,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27373,
            "output_tokens": 129,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27859,
            "output_tokens": 300,
            "reasoning_output_tokens": 108
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28492,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28917,
            "output_tokens": 166,
            "reasoning_output_tokens": 96
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29440,
            "output_tokens": 238,
            "reasoning_output_tokens": 170
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30035,
            "output_tokens": 86,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30479,
            "output_tokens": 221,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30842,
            "output_tokens": 68,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31079,
            "output_tokens": 294,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31574,
            "output_tokens": 482,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32078,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 31,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 12,
          "find": 1,
          "bash": 1,
          "exec": 11,
          "devices": 1,
          "device": 2,
          "write": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 252 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 8
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/workspace/.kinex/memos/robotwin.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/left_cartesian.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-37-20261116/workspace/.kinex/tools/left_cartesian.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 59,
        "observed_images": 9
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 102,
        "published_frames": 102,
        "decode_ok": true,
        "duration_s": 2.55,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "404ad34d99e76d0d5aada0a96632a7faf2c8773377a11a3eabda081d6eb44af0",
        "source_sha256": {
          "third_person": "fa57293584e6aa54265617a1ffcdc07c5977a625f0dba37dcb8ce97524398e2b",
          "left_wrist": "6eac94db15b15a26fd879b578e8549e7f8a911eb1675a578b302e2ffaed59d12",
          "right_wrist": "cf55dd84b3fca5cea4eeca2a228324dea02a576bd226cbc162ee2be637823644"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "a5261ed8bfed",
          "task": "robotwin/place-mouse-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 102,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 102,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10080000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 102,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 102,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10080000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 102,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 102,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10080000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": false
    },
    {
      "id": "kinex-pose-film-39-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/39",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "39",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 10.64,
        "step_limit": 400,
        "steps": 266,
        "success": true,
        "valid": true
      },
      "steps": 266,
      "simulation_time_s": 10.64,
      "wall_time_s": 843.32093,
      "timing": {
        "simulation_time_s": 10.64,
        "wall_time_s": 843.32093,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use one arm to pick the palm-sized black stapler for the black screen electronicscale",
      "native_instruction": "Use one arm to pick the palm-sized black stapler for the black screen electronicscale",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.940927789319054,
        "cache_read_input_tokens": 776320,
        "cache_reported_input_tokens": 825058,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 825058,
        "cached_input_tokens": 776320,
        "cost_usd": null,
        "fresh_input_tokens": 48738,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 825058,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 776320,
        "known_fresh_input_tokens": 48738,
        "known_input_tokens": 825058,
        "known_output_tokens": 4955,
        "known_reasoning_output_tokens": 484,
        "output_tokens": 4955,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 484,
        "reasoning_reported_output_tokens": 4955,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 30,
          "cache_write_input_tokens": 30,
          "cache_write_reported_input_tokens": 30,
          "cached_input_tokens": 30,
          "fresh_input_tokens": 30,
          "input_tokens": 30,
          "output_tokens": 30,
          "reasoning_output_tokens": 30,
          "reasoning_reported_output_tokens": 30
        },
        "request_attempts": 30,
        "response_count": 30,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 830013,
        "uncached_input_tokens": 48738,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11823,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11863,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11996,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 12046,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14960,
            "output_tokens": 43,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 15213,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15232,
            "input_tokens": 19469,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 23285,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23613,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 25219,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25410,
            "output_tokens": 113,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26929,
            "output_tokens": 271,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27670,
            "output_tokens": 1056,
            "reasoning_output_tokens": 171
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28751,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28913,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 30204,
            "output_tokens": 71,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30337,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30589,
            "output_tokens": 109,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 32151,
            "output_tokens": 136,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 33466,
            "output_tokens": 862,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34351,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34778,
            "output_tokens": 167,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35168,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35591,
            "output_tokens": 158,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36083,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36500,
            "output_tokens": 171,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36813,
            "output_tokens": 48,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36923,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37191,
            "output_tokens": 538,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37753,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 40,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 18,
          "find": 1,
          "bash": 1,
          "devices": 2,
          "exec": 10,
          "device": 1,
          "write": 3,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 266 rosbridge control frames; 24 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 6
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/manipulation.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/workspace/.kinex/memos/manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-39-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 73,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 24,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 108,
        "published_frames": 108,
        "decode_ok": true,
        "duration_s": 2.7,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "bedac99dfd612322c66d0baed7d1250a35a4a70ab2e42987de8dd20d43035a73",
        "source_sha256": {
          "third_person": "c964763f28cb846b597dd2396cb6f1f6fbaec98ce0efebff8878a82d8dabf9e3",
          "left_wrist": "bd9ceea2e218c019a85d44c31e8b23003a3324fcda96a4e0b1d635d0a6ac6d00",
          "right_wrist": "b37c936b1664d242d1a802f4b3fe8c30651c7d4b2a8eec5015d676ef0c7ec5c5"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "35935a1326c2",
          "task": "robotwin/place-object-scale-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 108,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 108,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 108,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 108,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 108,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 108,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-40-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/40",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "40",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.08,
        "step_limit": 400,
        "steps": 302,
        "success": true,
        "valid": true
      },
      "steps": 302,
      "simulation_time_s": 12.08,
      "wall_time_s": 940.138795,
      "timing": {
        "simulation_time_s": 12.08,
        "wall_time_s": 940.138795,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set the rectangular remotecontrol on the displaystand with hollow center.",
      "native_instruction": "Set the rectangular remotecontrol on the displaystand with hollow center.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9417221063961514,
        "cache_read_input_tokens": 860928,
        "cache_reported_input_tokens": 914206,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 914206,
        "cached_input_tokens": 860928,
        "cost_usd": null,
        "fresh_input_tokens": 53278,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 914206,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 860928,
        "known_fresh_input_tokens": 53278,
        "known_input_tokens": 914206,
        "known_output_tokens": 5730,
        "known_reasoning_output_tokens": 1076,
        "output_tokens": 5730,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1076,
        "reasoning_reported_output_tokens": 5730,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 919936,
        "uncached_input_tokens": 53278,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11816,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11856,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11989,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12037,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12386,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18130,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 20100,
            "output_tokens": 220,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 22606,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22798,
            "output_tokens": 175,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 25714,
            "output_tokens": 267,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26613,
            "output_tokens": 78,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27157,
            "output_tokens": 1001,
            "reasoning_output_tokens": 145
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28181,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28348,
            "output_tokens": 156,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 29245,
            "output_tokens": 126,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30111,
            "output_tokens": 70,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30391,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31420,
            "output_tokens": 903,
            "reasoning_output_tokens": 92
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32344,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32771,
            "output_tokens": 68,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32890,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 33028,
            "output_tokens": 89,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33135,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33215,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33479,
            "output_tokens": 220,
            "reasoning_output_tokens": 133
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 34034,
            "output_tokens": 345,
            "reasoning_output_tokens": 229
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34714,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35135,
            "output_tokens": 89,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35242,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35291,
            "output_tokens": 72,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35623,
            "output_tokens": 222,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 35872,
            "output_tokens": 638,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36535,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 47,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 17,
          "find": 4,
          "exec": 10,
          "devices": 3,
          "device": 6,
          "write": 3,
          "relative_pose": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 302 rosbridge control frames; 36 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 4
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-40-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 83,
        "observed_images": 13
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 36,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 122,
        "published_frames": 122,
        "decode_ok": true,
        "duration_s": 3.05,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "9390dd69172c33b3260938931959f0f9200714f4e41440f9a60c74a55c9be573",
        "source_sha256": {
          "third_person": "3975bf853edc57d89d52d2c2ae2b38bfab6ed6a8ead918ce7cf6939066b54d98",
          "left_wrist": "3487b7490b3a4c953d739a203121f3e73dc268cc4f70d0a07f6eb3589050dae6",
          "right_wrist": "24ad2b82eddcf6496c797bb74ed2b8d7b1faceee4c96c5d715e5f02d3c7bc055"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "568f5a6d3343",
          "task": "robotwin/place-object-stand-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 122,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 122,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12080000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 122,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 122,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12080000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 122,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 122,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12080000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-43-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/43",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "43",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 3.72,
        "step_limit": 400,
        "steps": 93,
        "success": true,
        "valid": true
      },
      "steps": 93,
      "simulation_time_s": 3.72,
      "wall_time_s": 289.636219,
      "timing": {
        "simulation_time_s": 3.72,
        "wall_time_s": 289.636219,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to press the shiny red plastic stapler.",
      "native_instruction": "Use the left arm to press the shiny red plastic stapler.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9554152765754638,
        "cache_read_input_tokens": 491136,
        "cache_reported_input_tokens": 514055,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 514055,
        "cached_input_tokens": 491136,
        "cost_usd": null,
        "fresh_input_tokens": 22919,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 514055,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 491136,
        "known_fresh_input_tokens": 22919,
        "known_input_tokens": 514055,
        "known_output_tokens": 3708,
        "known_reasoning_output_tokens": 660,
        "output_tokens": 3708,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 660,
        "reasoning_reported_output_tokens": 3708,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 24,
          "cache_write_input_tokens": 24,
          "cache_write_reported_input_tokens": 24,
          "cached_input_tokens": 24,
          "fresh_input_tokens": 24,
          "input_tokens": 24,
          "output_tokens": 24,
          "reasoning_output_tokens": 24,
          "reasoning_reported_output_tokens": 24
        },
        "request_attempts": 24,
        "response_count": 24,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 517763,
        "uncached_input_tokens": 22919,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11814,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11854,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11987,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12037,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12330,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 12517,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 16773,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 18743,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 19313,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 19477,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 19666,
            "output_tokens": 196,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 21940,
            "output_tokens": 460,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23977,
            "output_tokens": 310,
            "reasoning_output_tokens": 231
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25276,
            "output_tokens": 811,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26109,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26272,
            "output_tokens": 102,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26630,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27389,
            "output_tokens": 125,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27656,
            "output_tokens": 73,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 27751,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28212,
            "output_tokens": 60,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28381,
            "output_tokens": 345,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28754,
            "output_tokens": 419,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29197,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 31,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 13,
          "bash": 2,
          "exec": 6,
          "devices": 2,
          "device": 1,
          "write": 2,
          "device_stop": 1,
          "relative_pose": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 93 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 3
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/press-stapler.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/workspace/.kinex/memos/press-stapler.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/left_tcp_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-43-20261116/workspace/.kinex/tools/left_tcp_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 58,
        "observed_images": 4
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 38,
        "published_frames": 38,
        "decode_ok": true,
        "duration_s": 0.95,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "29f45c7454bfdf874014fd529c30f796af2a515f6f495666a3dc6379c7b6bfb1",
        "source_sha256": {
          "third_person": "3405eb254271614fecd752295f5288d1def827ccb0f12fef428908df13441df8",
          "left_wrist": "96bb62c20c5ee970549045d440cf62fc37575bee8332856a62fef2e0a372ae70",
          "right_wrist": "64f231ffda2cb42c6391ab4f0699ec3d1b11bb0fba48b8e8ad8b442d791cb81c"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "133b0c6902b1",
          "task": "robotwin/press-stapler-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 39,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 38,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3720000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 39,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 38,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3720000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 39,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 38,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3720000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": false
    },
    {
      "id": "kinex-pose-film-45-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/45",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "45",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 5.96,
        "step_limit": 700,
        "steps": 149,
        "success": true,
        "valid": true
      },
      "steps": 149,
      "simulation_time_s": 5.96,
      "wall_time_s": 450.128992,
      "timing": {
        "simulation_time_s": 5.96,
        "wall_time_s": 450.128992,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the bottle with white cap with the left arm and shake.",
      "native_instruction": "Pick up the bottle with white cap with the left arm and shake.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9494888822148808,
        "cache_read_input_tokens": 682880,
        "cache_reported_input_tokens": 719208,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 719208,
        "cached_input_tokens": 682880,
        "cost_usd": null,
        "fresh_input_tokens": 36328,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 719208,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 682880,
        "known_fresh_input_tokens": 36328,
        "known_input_tokens": 719208,
        "known_output_tokens": 4552,
        "known_reasoning_output_tokens": 496,
        "output_tokens": 4552,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 496,
        "reasoning_reported_output_tokens": 4552,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 34,
          "cache_write_input_tokens": 34,
          "cache_write_reported_input_tokens": 34,
          "cached_input_tokens": 34,
          "fresh_input_tokens": 34,
          "input_tokens": 34,
          "output_tokens": 34,
          "reasoning_output_tokens": 34,
          "reasoning_reported_output_tokens": 34
        },
        "request_attempts": 34,
        "response_count": 34,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 723760,
        "uncached_input_tokens": 36328,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11812,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11852,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11985,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12329,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12480,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 13294,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13184,
            "input_tokens": 14120,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13952,
            "input_tokens": 14583,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14464,
            "input_tokens": 15375,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 16599,
            "output_tokens": 76,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 17883,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18453,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 18617,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 19073,
            "output_tokens": 148,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 19431,
            "output_tokens": 305,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20131,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20539,
            "output_tokens": 305,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21940,
            "output_tokens": 53,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22459,
            "output_tokens": 211,
            "reasoning_output_tokens": 169
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 23659,
            "output_tokens": 1077,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24757,
            "output_tokens": 136,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 24959,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25710,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25796,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26150,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26566,
            "output_tokens": 146,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27040,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27521,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28340,
            "output_tokens": 112,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28594,
            "output_tokens": 65,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28681,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29130,
            "output_tokens": 259,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29416,
            "output_tokens": 495,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 29934,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 33,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 12,
          "bash": 1,
          "exec": 11,
          "devices": 3,
          "write": 2,
          "edit": 2,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 149 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 7
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/shake-bottle.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/workspace/.kinex/memos/shake-bottle.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-45-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 70,
        "observed_images": 5
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 61,
        "published_frames": 61,
        "decode_ok": true,
        "duration_s": 1.525,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "74be69b8e20dd973c65e938a0a44512ed11464f6b70fe5a4c4343ff5e10abac9",
        "source_sha256": {
          "third_person": "d146167750b75062b433ce128ff4faef6e41664d954f29a9b228c6a528b97141",
          "left_wrist": "60154466c775a99d17d9934a5a7a700e2c5c0167a893814fc8df4f9c5130cd90",
          "right_wrist": "2929e1794fb59bf951bc82b66ffebdabbc35ebfe924a42e7c387f19eadaf5541"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "969b6cb47c4f",
          "task": "robotwin/shake-bottle-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 61,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 61,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 5960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 61,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 61,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 5960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 61,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 61,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 5960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": false
    },
    {
      "id": "kinex-pose-film-38-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/38",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "38",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 27.96,
        "step_limit": 700,
        "steps": 699,
        "success": false,
        "valid": true
      },
      "steps": 699,
      "simulation_time_s": 27.96,
      "wall_time_s": 1864.473818,
      "timing": {
        "simulation_time_s": 27.96,
        "wall_time_s": 1864.473818,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the toycar with rounded edges and place into the basket with rounded edges.",
      "native_instruction": "Grab the toycar with rounded edges and place into the basket with rounded edges.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.981291015269898,
        "cache_read_input_tokens": 2110336,
        "cache_reported_input_tokens": 2150571,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2150571,
        "cached_input_tokens": 2110336,
        "cost_usd": null,
        "fresh_input_tokens": 40235,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2150571,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2110336,
        "known_fresh_input_tokens": 40235,
        "known_input_tokens": 2150571,
        "known_output_tokens": 13367,
        "known_reasoning_output_tokens": 4966,
        "output_tokens": 13367,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 4966,
        "reasoning_reported_output_tokens": 13367,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 59,
          "cache_write_input_tokens": 59,
          "cache_write_reported_input_tokens": 59,
          "cached_input_tokens": 59,
          "fresh_input_tokens": 59,
          "input_tokens": 59,
          "output_tokens": 59,
          "reasoning_output_tokens": 59,
          "reasoning_reported_output_tokens": 59
        },
        "request_attempts": 59,
        "response_count": 59,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2163938,
        "uncached_input_tokens": 40235,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11859,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11992,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12040,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13384,
            "output_tokens": 186,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13440,
            "input_tokens": 16126,
            "output_tokens": 234,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16256,
            "input_tokens": 21545,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 21873,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 23273,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23463,
            "output_tokens": 68,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23900,
            "output_tokens": 273,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24796,
            "output_tokens": 58,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25320,
            "output_tokens": 1134,
            "reasoning_output_tokens": 191
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26476,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26620,
            "output_tokens": 106,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27501,
            "output_tokens": 144,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28420,
            "output_tokens": 121,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 29738,
            "output_tokens": 1243,
            "reasoning_output_tokens": 188
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31002,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31577,
            "output_tokens": 183,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31982,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32757,
            "output_tokens": 54,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33990,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34803,
            "output_tokens": 162,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 34992,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35780,
            "output_tokens": 182,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 35989,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36558,
            "output_tokens": 233,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37651,
            "output_tokens": 136,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38266,
            "output_tokens": 68,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38356,
            "output_tokens": 299,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 39102,
            "output_tokens": 436,
            "reasoning_output_tokens": 348
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39603,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40166,
            "output_tokens": 142,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40788,
            "output_tokens": 359,
            "reasoning_output_tokens": 273
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41628,
            "output_tokens": 75,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41725,
            "output_tokens": 304,
            "reasoning_output_tokens": 145
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42251,
            "output_tokens": 333,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 43049,
            "output_tokens": 560,
            "reasoning_output_tokens": 472
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 44096,
            "output_tokens": 163,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44426,
            "output_tokens": 289,
            "reasoning_output_tokens": 153
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 44742,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45317,
            "output_tokens": 69,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45408,
            "output_tokens": 394,
            "reasoning_output_tokens": 295
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 46283,
            "output_tokens": 1155,
            "reasoning_output_tokens": 947
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47832,
            "output_tokens": 469,
            "reasoning_output_tokens": 328
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48128,
            "input_tokens": 48328,
            "output_tokens": 252,
            "reasoning_output_tokens": 153
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 48703,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48640,
            "input_tokens": 49289,
            "output_tokens": 125,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49280,
            "input_tokens": 49899,
            "output_tokens": 368,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50176,
            "input_tokens": 50685,
            "output_tokens": 172,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50688,
            "input_tokens": 50988,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50944,
            "input_tokens": 51548,
            "output_tokens": 310,
            "reasoning_output_tokens": 221
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51712,
            "input_tokens": 52643,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52480,
            "input_tokens": 52735,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52608,
            "input_tokens": 53150,
            "output_tokens": 278,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53248,
            "input_tokens": 53649,
            "output_tokens": 316,
            "reasoning_output_tokens": 156
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53760,
            "input_tokens": 53992,
            "output_tokens": 682,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54528,
            "input_tokens": 54698,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 71,
        "model_tool_calls_by_name": {
          "ls": 5,
          "read": 18,
          "find": 1,
          "devices": 2,
          "exec": 31,
          "device": 1,
          "write": 3,
          "edit": 5,
          "device_stop": 4,
          "grep": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 699 rosbridge control frames; 42 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 22
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin_basket.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/workspace/.kinex/memos/robotwin_basket.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-38-20261116/workspace/.kinex/tools/progvla_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 133,
        "observed_images": 34
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 42,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 281,
        "published_frames": 281,
        "decode_ok": true,
        "duration_s": 7.025,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "888c89cf63028d274eee453b674ad514a9cb186857851dbf421b839311783cb3",
        "source_sha256": {
          "third_person": "26812ec00d972441f367ccb0046f05eccd8f35b42f343e4b17a26f6c365e228f",
          "left_wrist": "a9a97ccfb789d0efe4d0dbebef042f3ed7e5ae3081b59189e296aac050faec74",
          "right_wrist": "ce4605cc6e8e56fa882109f7f363f4d8e1e9f08714ce78a1987cb4cd3788ec47"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "13b736267f24",
          "task": "robotwin/place-object-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 281,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 281,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 281,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 281,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 281,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 281,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-41-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/41",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "41",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 13.040000000000001,
        "step_limit": 400,
        "steps": 326,
        "success": true,
        "valid": true
      },
      "steps": 326,
      "simulation_time_s": 13.040000000000001,
      "wall_time_s": 1212.175818,
      "timing": {
        "simulation_time_s": 13.040000000000001,
        "wall_time_s": 1212.175818,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set the phone with shiny black back on the L-shaped smooth brown phonestand.",
      "native_instruction": "Set the phone with shiny black back on the L-shaped smooth brown phonestand.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9225494298667665,
        "cache_read_input_tokens": 956800,
        "cache_reported_input_tokens": 1037126,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1037126,
        "cached_input_tokens": 956800,
        "cost_usd": null,
        "fresh_input_tokens": 80326,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1037126,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 956800,
        "known_fresh_input_tokens": 80326,
        "known_input_tokens": 1037126,
        "known_output_tokens": 7359,
        "known_reasoning_output_tokens": 1650,
        "output_tokens": 7359,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1650,
        "reasoning_reported_output_tokens": 7359,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 37,
          "cache_write_input_tokens": 37,
          "cache_write_reported_input_tokens": 37,
          "cached_input_tokens": 37,
          "fresh_input_tokens": 37,
          "input_tokens": 37,
          "output_tokens": 37,
          "reasoning_output_tokens": 37,
          "reasoning_reported_output_tokens": 37
        },
        "request_attempts": 37,
        "response_count": 37,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1044485,
        "uncached_input_tokens": 80326,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11817,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11857,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11990,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12040,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 13068,
            "output_tokens": 123,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13056,
            "input_tokens": 13402,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13440,
            "input_tokens": 19146,
            "output_tokens": 114,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 21329,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21657,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 23057,
            "output_tokens": 130,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23583,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23991,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24540,
            "output_tokens": 356,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 25361,
            "output_tokens": 88,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25915,
            "output_tokens": 1089,
            "reasoning_output_tokens": 173
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27027,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27191,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 28267,
            "output_tokens": 151,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 29421,
            "output_tokens": 64,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29695,
            "output_tokens": 583,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30299,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 31000,
            "output_tokens": 182,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31804,
            "output_tokens": 60,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 31886,
            "output_tokens": 174,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32327,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 33039,
            "output_tokens": 389,
            "reasoning_output_tokens": 113
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33903,
            "output_tokens": 599,
            "reasoning_output_tokens": 413
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 35442,
            "output_tokens": 292,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 36176,
            "output_tokens": 194,
            "reasoning_output_tokens": 106
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36989,
            "output_tokens": 197,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 37807,
            "output_tokens": 278,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 39157,
            "output_tokens": 215,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39984,
            "output_tokens": 241,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40371,
            "output_tokens": 52,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 40485,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40677,
            "output_tokens": 726,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 41426,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 48,
        "model_tool_calls_by_name": {
          "ls": 5,
          "read": 15,
          "find": 2,
          "devices": 1,
          "exec": 19,
          "write": 3,
          "device_stop": 2,
          "device": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 326 rosbridge control frames; 43 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 8
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/place_phone_stand.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/workspace/.kinex/memos/place_phone_stand.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-41-20261116/workspace/.kinex/tools/tcp_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 88,
        "observed_images": 17
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 43,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 132,
        "published_frames": 132,
        "decode_ok": true,
        "duration_s": 3.3,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "4281a5a35747521885452e2a3772a0e4f7db5f568af124e4506ec7ca4b9aa36e",
        "source_sha256": {
          "third_person": "0eac09b43bddf03c9402a83f5890bdcdd87c0c401e9351a5c51c3bfd83a8fa72",
          "left_wrist": "557718a8acd7cb2403d8ca1c7cfdfdcfeab8dff941156265ea697cdfbdd59c13",
          "right_wrist": "134b06a342eae8956e2199f71017a55f22977a0934b7db57bea638a1982956fc"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "d008ffb864fd",
          "task": "robotwin/place-phone-stand-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 132,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 132,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13040000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 132,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 132,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13040000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 132,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 132,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13040000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-42-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/42",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "42",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 13.76,
        "step_limit": 500,
        "steps": 344,
        "success": false,
        "valid": true
      },
      "steps": 344,
      "simulation_time_s": 13.76,
      "wall_time_s": 1280.07046,
      "timing": {
        "simulation_time_s": 13.76,
        "wall_time_s": 1280.07046,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the casual brown shoe off the table and drop it on the mat",
      "native_instruction": "Grab the casual brown shoe off the table and drop it on the mat",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9543973151105899,
        "cache_read_input_tokens": 590080,
        "cache_reported_input_tokens": 618275,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 618275,
        "cached_input_tokens": 590080,
        "cost_usd": null,
        "fresh_input_tokens": 28195,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 618275,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 590080,
        "known_fresh_input_tokens": 28195,
        "known_input_tokens": 618275,
        "known_output_tokens": 4538,
        "known_reasoning_output_tokens": 654,
        "output_tokens": 4538,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 654,
        "reasoning_reported_output_tokens": 4538,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 23,
          "cache_write_input_tokens": 23,
          "cache_write_reported_input_tokens": 23,
          "cached_input_tokens": 23,
          "fresh_input_tokens": 23,
          "input_tokens": 23,
          "output_tokens": 23,
          "reasoning_output_tokens": 23,
          "reasoning_reported_output_tokens": 23
        },
        "request_attempts": 23,
        "response_count": 23,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 622813,
        "uncached_input_tokens": 28195,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11813,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11853,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11986,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 14726,
            "output_tokens": 267,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 19479,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 22712,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23282,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23610,
            "output_tokens": 161,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24167,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24575,
            "output_tokens": 162,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 27151,
            "output_tokens": 380,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27927,
            "output_tokens": 301,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28647,
            "output_tokens": 107,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29220,
            "output_tokens": 1006,
            "reasoning_output_tokens": 138
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30251,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30414,
            "output_tokens": 164,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 32533,
            "output_tokens": 136,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 34295,
            "output_tokens": 210,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 35963,
            "output_tokens": 139,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 37707,
            "output_tokens": 250,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38224,
            "output_tokens": 372,
            "reasoning_output_tokens": 150
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38627,
            "output_tokens": 460,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39113,
            "output_tokens": 15,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 30,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 10,
          "find": 1,
          "exec": 11,
          "devices": 1,
          "write": 2,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 344 rosbridge control frames; 81 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 4,
          "direct_motion_exec_calls": 1
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-shoe.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/workspace/.kinex/memos/robotwin-shoe.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-42-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 56,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 81,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 139,
        "published_frames": 139,
        "decode_ok": true,
        "duration_s": 3.475,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "96e8f47729f6b59fbd2ce508c7cf26b4bcccb72c21ab846c4bc274856aaefdad",
        "source_sha256": {
          "third_person": "cbd2d5f54662ec418e224c3361a4211e8cf3ac6ee200c551f7a5b1ea8ebc558f",
          "left_wrist": "0095554cf59af39eeca3e36ca415c43c6d3fbff4e78ef01953a42e3d05237552",
          "right_wrist": "9306c2cd5948cb461a151bfa547ca4e0de42397c97435b557c716199028fcfbb"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "6d8792702331",
          "task": "robotwin/place-shoe-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 139,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 139,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13760000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 139,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 139,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13760000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 139,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 139,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13760000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-44-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/44",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "44",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 13.64,
        "step_limit": 400,
        "steps": 341,
        "success": true,
        "valid": true
      },
      "steps": 341,
      "simulation_time_s": 13.64,
      "wall_time_s": 1172.966522,
      "timing": {
        "simulation_time_s": 13.64,
        "wall_time_s": 1172.966522,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Catch the tabletop paymentsign, pick it up and rotate it, then set it back down.",
      "native_instruction": "Catch the tabletop paymentsign, pick it up and rotate it, then set it back down.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9704694705074058,
        "cache_read_input_tokens": 920960,
        "cache_reported_input_tokens": 948984,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 948984,
        "cached_input_tokens": 920960,
        "cost_usd": null,
        "fresh_input_tokens": 28024,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 948984,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 920960,
        "known_fresh_input_tokens": 28024,
        "known_input_tokens": 948984,
        "known_output_tokens": 6583,
        "known_reasoning_output_tokens": 1218,
        "output_tokens": 6583,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1218,
        "reasoning_reported_output_tokens": 6583,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 35,
          "cache_write_input_tokens": 35,
          "cache_write_reported_input_tokens": 35,
          "cached_input_tokens": 35,
          "fresh_input_tokens": 35,
          "input_tokens": 35,
          "output_tokens": 35,
          "reasoning_output_tokens": 35,
          "reasoning_reported_output_tokens": 35
        },
        "request_attempts": 35,
        "response_count": 35,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 955567,
        "uncached_input_tokens": 28024,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11818,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11858,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11991,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12039,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 13298,
            "output_tokens": 165,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13312,
            "input_tokens": 16768,
            "output_tokens": 202,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16768,
            "input_tokens": 21370,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 21698,
            "output_tokens": 156,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22250,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22658,
            "output_tokens": 155,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 24544,
            "output_tokens": 291,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25232,
            "output_tokens": 299,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25853,
            "output_tokens": 137,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26456,
            "output_tokens": 971,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27450,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27613,
            "output_tokens": 122,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28363,
            "output_tokens": 87,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28660,
            "output_tokens": 53,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28775,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29994,
            "output_tokens": 919,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 30934,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31328,
            "output_tokens": 145,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31786,
            "output_tokens": 218,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32325,
            "output_tokens": 105,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32754,
            "output_tokens": 506,
            "reasoning_output_tokens": 429
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33328,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33728,
            "output_tokens": 189,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34231,
            "output_tokens": 134,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34665,
            "output_tokens": 172,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35038,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35443,
            "output_tokens": 280,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35865,
            "output_tokens": 55,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 35982,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36175,
            "output_tokens": 516,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 36714,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 43,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 16,
          "find": 1,
          "devices": 1,
          "exec": 17,
          "write": 3,
          "device_stop": 2,
          "device": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 341 rosbridge control frames; 15 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 8
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/rotate-qrcode.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/workspace/.kinex/memos/rotate-qrcode.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-44-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 81,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 15,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 138,
        "published_frames": 138,
        "decode_ok": true,
        "duration_s": 3.45,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "93c653db97b846adf7df99e483c9e4d2cce1f06d097d3268979c208766db04d7",
        "source_sha256": {
          "third_person": "6b6e74dcaec13dade02795d434add79d42f9fd7034d232ceef980d3f99d25828",
          "left_wrist": "7270009df44d30beb126b08f357ad60cf66d504d1a6b7680ae2db666c77a3204",
          "right_wrist": "77848d108e42eca0cacd75325d8449c230f6b04e133bee6962f6972966d5aa9e"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "99befa29f5b7",
          "task": "robotwin/rotate-qrcode-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 138,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 138,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 138,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-46-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/46",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "46",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 11.24,
        "step_limit": 700,
        "steps": 281,
        "success": true,
        "valid": true
      },
      "steps": 281,
      "simulation_time_s": 11.24,
      "wall_time_s": 983.045596,
      "timing": {
        "simulation_time_s": 11.24,
        "wall_time_s": 983.045596,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the bottle with white cap and shake it horizontally.",
      "native_instruction": "Grab the bottle with white cap and shake it horizontally.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9535599787369279,
        "cache_read_input_tokens": 751616,
        "cache_reported_input_tokens": 788221,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 788221,
        "cached_input_tokens": 751616,
        "cost_usd": null,
        "fresh_input_tokens": 36605,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 788221,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 751616,
        "known_fresh_input_tokens": 36605,
        "known_input_tokens": 788221,
        "known_output_tokens": 5041,
        "known_reasoning_output_tokens": 486,
        "output_tokens": 5041,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 486,
        "reasoning_reported_output_tokens": 5041,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 28,
          "cache_write_input_tokens": 28,
          "cache_write_reported_input_tokens": 28,
          "cached_input_tokens": 28,
          "fresh_input_tokens": 28,
          "input_tokens": 28,
          "output_tokens": 28,
          "reasoning_output_tokens": 28,
          "reasoning_reported_output_tokens": 28
        },
        "request_attempts": 28,
        "response_count": 28,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 793262,
        "uncached_input_tokens": 36605,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11816,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11856,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11989,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15537,
            "output_tokens": 218,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15616,
            "input_tokens": 19096,
            "output_tokens": 202,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 23698,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24026,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 25426,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25617,
            "output_tokens": 109,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26765,
            "output_tokens": 279,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27439,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27847,
            "output_tokens": 274,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28652,
            "output_tokens": 89,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29207,
            "output_tokens": 961,
            "reasoning_output_tokens": 92
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30191,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30356,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31093,
            "output_tokens": 129,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 32419,
            "output_tokens": 120,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32997,
            "output_tokens": 908,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33926,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34437,
            "output_tokens": 117,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34980,
            "output_tokens": 151,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35351,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 36129,
            "output_tokens": 124,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36395,
            "output_tokens": 52,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36509,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36956,
            "output_tokens": 530,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37511,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 40,
        "model_tool_calls_by_name": {
          "ls": 4,
          "read": 17,
          "find": 1,
          "devices": 2,
          "exec": 10,
          "device": 1,
          "write": 3,
          "grep": 1,
          "device_stop": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 281 rosbridge control frames; 40 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 5
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-46-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 71,
        "observed_images": 7
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 40,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 114,
        "published_frames": 114,
        "decode_ok": true,
        "duration_s": 2.85,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "629ad110ae89d55f56705aca4ed37ba0fc97a27ee4d32f877440c8e316310c13",
        "source_sha256": {
          "third_person": "a95def452feda44bdbc47d80156783237c0442a222871ec2696d3671b0d16225",
          "left_wrist": "b49ad3d7e0e0ad4f2a0cbe6bde6aeb8769cb1b4e41d3bf861be901734b4622ec",
          "right_wrist": "c67749ad529db2148780011418b3cb7a2c88f752fe483e6ef6e56fa5045fc078"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "0b695c9167b6",
          "task": "robotwin/shake-bottle-horizontally-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 114,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 114,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11240000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 114,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 114,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11240000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 114,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 114,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11240000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-47-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/47",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "47",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 19.56,
        "step_limit": 800,
        "steps": 489,
        "success": true,
        "valid": true
      },
      "steps": 489,
      "simulation_time_s": 19.56,
      "wall_time_s": 1083.121537,
      "timing": {
        "simulation_time_s": 19.56,
        "wall_time_s": 1083.121537,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Bring red block and green block to the center, then position green block above red block.",
      "native_instruction": "Bring red block and green block to the center, then position green block above red block.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9685868738514738,
        "cache_read_input_tokens": 1230208,
        "cache_reported_input_tokens": 1270106,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1270106,
        "cached_input_tokens": 1230208,
        "cost_usd": null,
        "fresh_input_tokens": 39898,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1270106,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1230208,
        "known_fresh_input_tokens": 39898,
        "known_input_tokens": 1270106,
        "known_output_tokens": 6445,
        "known_reasoning_output_tokens": 951,
        "output_tokens": 6445,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 951,
        "reasoning_reported_output_tokens": 6445,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 42,
          "cache_write_input_tokens": 42,
          "cache_write_reported_input_tokens": 42,
          "cached_input_tokens": 42,
          "fresh_input_tokens": 42,
          "input_tokens": 42,
          "output_tokens": 42,
          "reasoning_output_tokens": 42,
          "reasoning_reported_output_tokens": 42
        },
        "request_attempts": 42,
        "response_count": 42,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1276551,
        "uncached_input_tokens": 39898,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11819,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11952,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12002,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 12295,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12446,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 18190,
            "output_tokens": 64,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18785,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 19015,
            "output_tokens": 159,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 19569,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 20239,
            "output_tokens": 157,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 21399,
            "output_tokens": 461,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22578,
            "output_tokens": 99,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 23143,
            "output_tokens": 87,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 24409,
            "output_tokens": 1529,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25959,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26529,
            "output_tokens": 169,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27368,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27461,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28145,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28821,
            "output_tokens": 61,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29459,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30138,
            "output_tokens": 129,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 31019,
            "output_tokens": 81,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 33980,
            "output_tokens": 111,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34118,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34978,
            "output_tokens": 87,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35857,
            "output_tokens": 67,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36619,
            "output_tokens": 547,
            "reasoning_output_tokens": 189
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 38039,
            "output_tokens": 84,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38716,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39388,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 40002,
            "output_tokens": 87,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40686,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 41361,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 42036,
            "output_tokens": 44,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41856,
            "input_tokens": 42646,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43320,
            "output_tokens": 108,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 44027,
            "output_tokens": 139,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44763,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 44950,
            "output_tokens": 71,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45553,
            "output_tokens": 752,
            "reasoning_output_tokens": 102
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46327,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 48,
        "model_tool_calls_by_name": {
          "read": 15,
          "ls": 1,
          "bash": 1,
          "devices": 1,
          "exec": 25,
          "write": 2,
          "edit": 2,
          "grep": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 489 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 21
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack_blocks.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/workspace/.kinex/memos/stack_blocks.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/stack_blocks.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-47-20261116/workspace/.kinex/tools/stack_blocks.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 93,
        "observed_images": 17
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 197,
        "published_frames": 197,
        "decode_ok": true,
        "duration_s": 4.925,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "aee36830eafe958dfb3d66ca609a4dd71a4c327c2437ecb1f43ab38348fd8bbd",
        "source_sha256": {
          "third_person": "be4727df0d784028f7255346218507892b8e4cbc7105696411e69a1b457679d1",
          "left_wrist": "09d9d8013630cf01c9ec3c6454b5c3b6afd23fbff642b53f22a0e59f08f25987",
          "right_wrist": "35927994cc8b9e56f6f0c04b7ebb3d995ae16f5035c1de41c925a1f6f405f1d0"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "8219347d4e23",
          "task": "robotwin/stack-blocks-two-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 197,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 197,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 197,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 197,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 197,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 197,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": false
    },
    {
      "id": "kinex-pose-film-48-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/48",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "48",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 27.96,
        "step_limit": 900,
        "steps": 699,
        "success": true,
        "valid": true
      },
      "steps": 699,
      "simulation_time_s": 27.96,
      "wall_time_s": 1424.278459,
      "timing": {
        "simulation_time_s": 27.96,
        "wall_time_s": 1424.278459,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Stack the small smooth brown-rimmed bowl directly over the smooth bowl with glossy finish.",
      "native_instruction": "Stack the small smooth brown-rimmed bowl directly over the smooth bowl with glossy finish.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9639525651576509,
        "cache_read_input_tokens": 1063552,
        "cache_reported_input_tokens": 1103324,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1103324,
        "cached_input_tokens": 1063552,
        "cost_usd": null,
        "fresh_input_tokens": 39772,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1103324,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1063552,
        "known_fresh_input_tokens": 39772,
        "known_input_tokens": 1103324,
        "known_output_tokens": 7199,
        "known_reasoning_output_tokens": 1806,
        "output_tokens": 7199,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1806,
        "reasoning_reported_output_tokens": 7199,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 36,
          "cache_write_input_tokens": 36,
          "cache_write_reported_input_tokens": 36,
          "cached_input_tokens": 36,
          "fresh_input_tokens": 36,
          "input_tokens": 36,
          "output_tokens": 36,
          "reasoning_output_tokens": 36,
          "reasoning_reported_output_tokens": 36
        },
        "request_attempts": 36,
        "response_count": 36,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1110523,
        "uncached_input_tokens": 39772,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11823,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11871,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12002,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12295,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18184,
            "output_tokens": 188,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 22350,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22678,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22868,
            "output_tokens": 90,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23997,
            "output_tokens": 302,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24694,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25102,
            "output_tokens": 296,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 27282,
            "output_tokens": 129,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 28309,
            "output_tokens": 251,
            "reasoning_output_tokens": 211
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 29026,
            "output_tokens": 1189,
            "reasoning_output_tokens": 295
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30238,
            "output_tokens": 130,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 31353,
            "output_tokens": 343,
            "reasoning_output_tokens": 210
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32683,
            "output_tokens": 66,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32959,
            "output_tokens": 861,
            "reasoning_output_tokens": 148
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 33841,
            "output_tokens": 121,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34329,
            "output_tokens": 110,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34810,
            "output_tokens": 82,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35264,
            "output_tokens": 131,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35764,
            "output_tokens": 117,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36252,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36699,
            "output_tokens": 87,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37156,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37604,
            "output_tokens": 349,
            "reasoning_output_tokens": 140
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38515,
            "output_tokens": 222,
            "reasoning_output_tokens": 140
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 39108,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39558,
            "output_tokens": 105,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40032,
            "output_tokens": 273,
            "reasoning_output_tokens": 218
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40447,
            "output_tokens": 72,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40581,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40775,
            "output_tokens": 367,
            "reasoning_output_tokens": 167
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41171,
            "output_tokens": 510,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41704,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 45,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 16,
          "bash": 1,
          "exec": 19,
          "devices": 1,
          "device": 2,
          "write": 3,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 699 rosbridge control frames; 90 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 2,
          "direct_motion_exec_calls": 12
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/bowl_stacking.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/workspace/.kinex/memos/bowl_stacking.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-48-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 84,
        "observed_images": 20
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 90,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 281,
        "published_frames": 281,
        "decode_ok": true,
        "duration_s": 7.025,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f592cbd72ab5849fb2febcced62afd79d65d43f06a7b254bc97c436a4e6c8caf",
        "source_sha256": {
          "third_person": "8869a97d618b0f339e762fac8d0a566a6039015492e1199b3fecb816d0727e5a",
          "left_wrist": "1021be77fd97f51578d288bb83d7580d3d634c78c334ae3d712d0f7896ddf7d4",
          "right_wrist": "521b5902ace4e506997b817912a491ee0a91051a066209088a510249b7e79ae4"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "c4a63604058e",
          "task": "robotwin/stack-bowls-two-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 281,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 281,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 281,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 281,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 281,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 281,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-49-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/49",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "49",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 11.120000000000001,
        "step_limit": 400,
        "steps": 278,
        "success": false,
        "valid": true
      },
      "steps": 278,
      "simulation_time_s": 11.120000000000001,
      "wall_time_s": 926.151275,
      "timing": {
        "simulation_time_s": 11.120000000000001,
        "wall_time_s": 926.151275,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Stamp Green using the block-shaped seal with angled edges",
      "native_instruction": "Stamp Green using the block-shaped seal with angled edges",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.932795126727531,
        "cache_read_input_tokens": 723072,
        "cache_reported_input_tokens": 775167,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 775167,
        "cached_input_tokens": 723072,
        "cost_usd": null,
        "fresh_input_tokens": 52095,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 775167,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 723072,
        "known_fresh_input_tokens": 52095,
        "known_input_tokens": 775167,
        "known_output_tokens": 5257,
        "known_reasoning_output_tokens": 958,
        "output_tokens": 5257,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 958,
        "reasoning_reported_output_tokens": 5257,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 28,
          "cache_write_input_tokens": 28,
          "cache_write_reported_input_tokens": 28,
          "cached_input_tokens": 28,
          "fresh_input_tokens": 28,
          "input_tokens": 28,
          "output_tokens": 28,
          "reasoning_output_tokens": 28,
          "reasoning_reported_output_tokens": 28
        },
        "request_attempts": 28,
        "response_count": 28,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 780424,
        "uncached_input_tokens": 52095,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11815,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11948,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11998,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12291,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 15525,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15488,
            "input_tokens": 15746,
            "output_tokens": 288,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 22199,
            "output_tokens": 109,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22527,
            "output_tokens": 169,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 25855,
            "output_tokens": 358,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 27925,
            "output_tokens": 945,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28893,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29514,
            "output_tokens": 67,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29791,
            "output_tokens": 91,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30348,
            "output_tokens": 641,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31010,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31476,
            "output_tokens": 104,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31973,
            "output_tokens": 169,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32330,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32808,
            "output_tokens": 160,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33357,
            "output_tokens": 254,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33897,
            "output_tokens": 201,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 34490,
            "output_tokens": 112,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34996,
            "output_tokens": 132,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35823,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35878,
            "output_tokens": 131,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36403,
            "output_tokens": 643,
            "reasoning_output_tokens": 236
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37069,
            "output_tokens": 184,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37282,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 39,
        "model_tool_calls_by_name": {
          "read": 16,
          "ls": 1,
          "bash": 2,
          "exec": 13,
          "devices": 1,
          "device": 1,
          "write": 3,
          "device_stop": 1,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 278 rosbridge control frames; 25 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 1,
          "direct_motion_exec_calls": 9
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stamp_seal.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/workspace/.kinex/memos/stamp_seal.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/progvla_subgoal.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/workspace/.kinex/tools/progvla_subgoal.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/tcp_move.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-49-20261116/workspace/.kinex/tools/tcp_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 70,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 25,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 112,
        "published_frames": 112,
        "decode_ok": true,
        "duration_s": 2.8,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "ab1dbbbd60dee4199d38bd75e4c3e789e9a8c81d023cb478175d3b03218af002",
        "source_sha256": {
          "third_person": "5a5a435d22b2a3f73588758b831948a724b7d2e97e8cbbcf435395aec6fbf035",
          "left_wrist": "96535aedd183c0d1f35634aaa3ebb13a91c93e95d74ef7ac117ce5d5832c4a5c",
          "right_wrist": "18ce64b6d662acf9a51044012030cc7707048ed0658cff80321d93f916071a19"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "b98d2956f5d9",
          "task": "robotwin/stamp-seal-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 113,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 112,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11120000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 113,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 112,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11120000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 113,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 112,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11120000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": true
    },
    {
      "id": "kinex-pose-film-50-20261116",
      "variant": "kinex-pose-film",
      "task_key": "task03/50",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "50",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "completed",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.4,
        "step_limit": 400,
        "steps": 385,
        "success": false,
        "valid": true
      },
      "steps": 385,
      "simulation_time_s": 15.4,
      "wall_time_s": 1005.175462,
      "timing": {
        "simulation_time_s": 15.4,
        "wall_time_s": 1005.175462,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Click the flat gray switch with the right arm",
      "native_instruction": "Click the flat gray switch with the right arm",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9403614841542961,
        "cache_read_input_tokens": 1092480,
        "cache_reported_input_tokens": 1161766,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1161766,
        "cached_input_tokens": 1092480,
        "cost_usd": null,
        "fresh_input_tokens": 69286,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1161766,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1092480,
        "known_fresh_input_tokens": 69286,
        "known_input_tokens": 1161766,
        "known_output_tokens": 8094,
        "known_reasoning_output_tokens": 2268,
        "output_tokens": 8094,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2268,
        "reasoning_reported_output_tokens": 8094,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 41,
          "cache_write_input_tokens": 41,
          "cache_write_reported_input_tokens": 41,
          "cached_input_tokens": 41,
          "fresh_input_tokens": 41,
          "input_tokens": 41,
          "output_tokens": 41,
          "reasoning_output_tokens": 41,
          "reasoning_reported_output_tokens": 41
        },
        "request_attempts": 41,
        "response_count": 41,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1169860,
        "uncached_input_tokens": 69286,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11813,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11853,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11986,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12036,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12329,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12792,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12800,
            "input_tokens": 16596,
            "output_tokens": 103,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 18466,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 18630,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 18819,
            "output_tokens": 167,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 20600,
            "output_tokens": 358,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 22959,
            "output_tokens": 126,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 23551,
            "output_tokens": 742,
            "reasoning_output_tokens": 96
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24689,
            "output_tokens": 150,
            "reasoning_output_tokens": 108
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 26018,
            "output_tokens": 337,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 27359,
            "output_tokens": 65,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 28119,
            "output_tokens": 245,
            "reasoning_output_tokens": 203
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 29353,
            "output_tokens": 893,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30267,
            "output_tokens": 115,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30442,
            "output_tokens": 50,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30514,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30998,
            "output_tokens": 186,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 31211,
            "output_tokens": 174,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31557,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32052,
            "output_tokens": 221,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 32300,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32858,
            "output_tokens": 172,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33468,
            "output_tokens": 155,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 34085,
            "output_tokens": 167,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 35253,
            "output_tokens": 724,
            "reasoning_output_tokens": 463
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36062,
            "output_tokens": 269,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36738,
            "output_tokens": 114,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37313,
            "output_tokens": 134,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37913,
            "output_tokens": 135,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 38483,
            "output_tokens": 516,
            "reasoning_output_tokens": 409
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 39439,
            "output_tokens": 49,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39506,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40077,
            "output_tokens": 301,
            "reasoning_output_tokens": 190
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40811,
            "output_tokens": 99,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 40932,
            "output_tokens": 565,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41519,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 46,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 14,
          "bash": 1,
          "exec": 19,
          "devices": 2,
          "device": 2,
          "write": 2,
          "device_stop": 2,
          "edit": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 385 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; Hard tasks 01-50.",
        "vla_contribution": {
          "progvla_exec_calls": 0,
          "direct_motion_exec_calls": 11
        },
        "policy_call_scope": "Completed policy calls count model requests, not task contribution. Motion executions are classified from the session tool inputs."
      },
      "links": {
        "video": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/presentation/video.mp4",
        "poster": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/transcript.json",
        "usage": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/usage.json",
        "analysis": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/analysis.json",
        "protocol": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/protocol.json",
        "native_goal": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/instruction.json",
        "verdict": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/episode.json",
        "media_check": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/artifact-index.json",
        "vla_io": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/evidence.zip",
        "timing": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/timing.json",
        "native_session": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/session.jsonl",
        "trace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/trace.json",
        "provider_audit": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/provider-requests.jsonl",
        "provenance": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
        "workspace": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/turn-switch.md",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/workspace/.kinex/memos/turn-switch.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/right_switch_motion.py",
          "file": "experiments/kinex-pose-film-hard-native-20261116/kinex-pose-film-50-20261116/workspace/.kinex/tools/right_switch_motion.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 90,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-film-model-device-20261116",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 155,
        "published_frames": 155,
        "decode_ok": true,
        "duration_s": 3.875,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "5a8cb100e74204520d1a73648a368ac45db4241dd2fe682c45c19ee0793f6739",
        "source_sha256": {
          "third_person": "f7b8b0d95da01f995c72f2230fa05c5af1e3827672bfc22f0a411372ce17ce36",
          "left_wrist": "49e565b6f1ce9700aa30c2c1622b832aefed74fabeef43a2e0e45aef43261392",
          "right_wrist": "3f8d728b858ac50a4454cbfc93d4183e85a7d780724effbb572d40c576210413"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "b307fcc4b4f5",
          "task": "robotwin/turn-switch-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 156,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 155,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15400000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 156,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 155,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15400000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 156,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 155,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15400000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-film-model-device-20261116",
      "campaign": "progvla-yz-film-hard50-20261116",
      "evaluation_date": "2026-11-16",
      "vla_model": "progVLA-yz-film",
      "vla_used": false
    }
  ],
  "attempt_history": [
    {
      "id": "hard-kinex-01",
      "variant": "kinex",
      "task_key": "task03/01",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "01",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 18.44,
        "step_limit": 800,
        "steps": 461,
        "success": true,
        "valid": true
      },
      "steps": 461,
      "simulation_time_s": 18.44,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 18.44,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "native_instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9715530148359874,
        "cache_read_input_tokens": 1827328,
        "cache_reported_input_tokens": 1880832,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1880832,
        "cached_input_tokens": 1827328,
        "cost_usd": null,
        "fresh_input_tokens": 53504,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1880832,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1827328,
        "known_fresh_input_tokens": 53504,
        "known_input_tokens": 1880832,
        "known_output_tokens": 9369,
        "known_reasoning_output_tokens": 2710,
        "output_tokens": 9369,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2710,
        "reasoning_reported_output_tokens": 9369,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 50,
          "cache_write_input_tokens": 50,
          "cache_write_reported_input_tokens": 50,
          "cached_input_tokens": 50,
          "fresh_input_tokens": 50,
          "input_tokens": 50,
          "output_tokens": 50,
          "reasoning_output_tokens": 50,
          "reasoning_reported_output_tokens": 50
        },
        "request_attempts": 50,
        "response_count": 50,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1890201,
        "uncached_input_tokens": 53504,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11644,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11684,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11817,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 15903,
            "output_tokens": 199,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 19009,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 22809,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 24056,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24286,
            "output_tokens": 146,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 25503,
            "output_tokens": 223,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26918,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 28137,
            "output_tokens": 1324,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29482,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30225,
            "output_tokens": 184,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30813,
            "output_tokens": 201,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31385,
            "output_tokens": 373,
            "reasoning_output_tokens": 245
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32486,
            "output_tokens": 103,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33315,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 34092,
            "output_tokens": 319,
            "reasoning_output_tokens": 248
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 35154,
            "output_tokens": 234,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 36100,
            "output_tokens": 87,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36903,
            "output_tokens": 86,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37011,
            "output_tokens": 70,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37776,
            "output_tokens": 111,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38011,
            "output_tokens": 335,
            "reasoning_output_tokens": 226
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 39050,
            "output_tokens": 63,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39861,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 39979,
            "output_tokens": 106,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40837,
            "output_tokens": 191,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41055,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41255,
            "output_tokens": 371,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41965,
            "output_tokens": 236,
            "reasoning_output_tokens": 158
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42922,
            "output_tokens": 388,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43337,
            "output_tokens": 338,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 44095,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44897,
            "output_tokens": 138,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45147,
            "output_tokens": 136,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 46006,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46777,
            "output_tokens": 82,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 47585,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47616,
            "input_tokens": 48383,
            "output_tokens": 276,
            "reasoning_output_tokens": 166
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48512,
            "input_tokens": 49600,
            "output_tokens": 191,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49664,
            "input_tokens": 49918,
            "output_tokens": 110,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49920,
            "input_tokens": 50742,
            "output_tokens": 138,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50816,
            "input_tokens": 51591,
            "output_tokens": 155,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51584,
            "input_tokens": 51935,
            "output_tokens": 106,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51840,
            "input_tokens": 52753,
            "output_tokens": 352,
            "reasoning_output_tokens": 304
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52992,
            "input_tokens": 53239,
            "output_tokens": 60,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53120,
            "input_tokens": 53979,
            "output_tokens": 340,
            "reasoning_output_tokens": 179
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54144,
            "input_tokens": 54346,
            "output_tokens": 691,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54912,
            "input_tokens": 55059,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 62
      },
      "analysis": {
        "observed": "Native success after 461 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-01/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-01/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-01/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-01/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-01/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-01/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-01/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-01/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-01/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-01/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-01/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-01/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-01/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-01/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-01/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-01/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-01/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-01/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/handover.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-01/workspace/.kinex/tools/handover.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/handover.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-01/workspace/.kinex/memos/handover.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 114,
        "observed_images": 18
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 186,
        "published_frames": 186,
        "decode_ok": true,
        "duration_s": 4.65,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "5bff2cf36aefe4adbe1c15104d0f6ca2530a48640ac98e998b5c01c1a5a10325",
        "source_sha256": {
          "third_person": "ee0c012b21e0a5ba15ec819b9d6afca6690aa52316240f9445fe43a29b57b55c",
          "left_wrist": "9d51c9160bfcd25f370c01de9b2371c618c512734913f0cb405df43144ca137e",
          "right_wrist": "bb958927beea3cedc2931e10f91650c8c68dfa034a35b10663b811e571689290"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "579ea0ec80da",
          "task": "robotwin/handover-block-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 186,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 186,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18440000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 186,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 186,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18440000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 186,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 186,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 18440000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-02",
      "variant": "kinex",
      "task_key": "task03/02",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "02",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 19.76,
        "step_limit": 600,
        "steps": 494,
        "success": false,
        "valid": true
      },
      "steps": 494,
      "simulation_time_s": 19.76,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 19.76,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "native_instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9620603621730383,
        "cache_read_input_tokens": 956288,
        "cache_reported_input_tokens": 994000,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 994000,
        "cached_input_tokens": 956288,
        "cost_usd": null,
        "fresh_input_tokens": 37712,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 994000,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 956288,
        "known_fresh_input_tokens": 37712,
        "known_input_tokens": 994000,
        "known_output_tokens": 6521,
        "known_reasoning_output_tokens": 1910,
        "output_tokens": 6521,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1910,
        "reasoning_reported_output_tokens": 6521,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 36,
          "cache_write_input_tokens": 36,
          "cache_write_reported_input_tokens": 36,
          "cached_input_tokens": 36,
          "fresh_input_tokens": 36,
          "input_tokens": 36,
          "output_tokens": 36,
          "reasoning_output_tokens": 36,
          "reasoning_reported_output_tokens": 36
        },
        "request_attempts": 36,
        "response_count": 36,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1000521,
        "uncached_input_tokens": 37712,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11639,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11674,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11805,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11853,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 13921,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13952,
            "input_tokens": 17754,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 20269,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 20499,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20690,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 21269,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21521,
            "output_tokens": 408,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22324,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22595,
            "output_tokens": 334,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 24753,
            "output_tokens": 217,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 28071,
            "output_tokens": 1215,
            "reasoning_output_tokens": 380
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29307,
            "output_tokens": 111,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30044,
            "output_tokens": 45,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30151,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30230,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30787,
            "output_tokens": 133,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31312,
            "output_tokens": 105,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31817,
            "output_tokens": 66,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32266,
            "output_tokens": 156,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32814,
            "output_tokens": 397,
            "reasoning_output_tokens": 242
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33609,
            "output_tokens": 395,
            "reasoning_output_tokens": 320
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34307,
            "output_tokens": 97,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34426,
            "output_tokens": 100,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34825,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35224,
            "output_tokens": 360,
            "reasoning_output_tokens": 144
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35942,
            "output_tokens": 229,
            "reasoning_output_tokens": 154
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36467,
            "output_tokens": 426,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37099,
            "output_tokens": 138,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37538,
            "output_tokens": 65,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37897,
            "output_tokens": 130,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38329,
            "output_tokens": 619,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 38972,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 43
      },
      "analysis": {
        "observed": "Native failure after 494 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-02/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-02/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-02/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-02/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-02/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-02/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-02/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-02/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-02/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-02/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-02/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-02/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-02/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-02/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-02/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-02/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-02/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-02/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/shoe_motion.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-02/workspace/.kinex/tools/shoe_motion.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-shoes.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-02/workspace/.kinex/memos/robotwin-shoes.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 81,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 199,
        "published_frames": 199,
        "decode_ok": true,
        "duration_s": 4.975,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "acfce615d3fa888b2180bb50fd58ff9b553c05e8815d7e94df8c077ab000f952",
        "source_sha256": {
          "third_person": "2ef63ada36f2ba988745c98c7ead93e26c1449b34a61948b46eda09b12bc8d30",
          "left_wrist": "84e26ddb6052951f1a86a68aff601e94ea2500b6b5bf96bff98b6a9397d83dbd",
          "right_wrist": "cfa3ca0bc4993386b251eb6f8c958facae21ef0e14320b0ec531954d2181dc66"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "20e4ecc51495",
          "task": "robotwin/place-dual-shoes-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 199,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 199,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19760000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 199,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 199,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19760000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 199,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 199,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19760000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-03",
      "variant": "kinex",
      "task_key": "task03/03",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "03",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 2
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 65.2,
        "step_limit": 1700,
        "steps": 1630,
        "success": false,
        "valid": true
      },
      "steps": 1630,
      "simulation_time_s": 65.2,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 65.2,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap in the dustbin the simple black garbage bin.",
      "native_instruction": "Place the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap in the dustbin the simple black garbage bin.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9679528720365247,
        "cache_read_input_tokens": 3432448,
        "cache_reported_input_tokens": 3546090,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 3546090,
        "cached_input_tokens": 3432448,
        "cost_usd": null,
        "fresh_input_tokens": 113642,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 3546090,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 3432448,
        "known_fresh_input_tokens": 113642,
        "known_input_tokens": 3546090,
        "known_output_tokens": 22616,
        "known_reasoning_output_tokens": 10026,
        "output_tokens": 22616,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 10026,
        "reasoning_reported_output_tokens": 22616,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 94,
          "cache_write_input_tokens": 94,
          "cache_write_reported_input_tokens": 94,
          "cached_input_tokens": 94,
          "fresh_input_tokens": 94,
          "input_tokens": 94,
          "output_tokens": 94,
          "reasoning_output_tokens": 94,
          "reasoning_reported_output_tokens": 94
        },
        "request_attempts": 94,
        "response_count": 94,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 3568706,
        "uncached_input_tokens": 113642,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11659,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11699,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11832,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11882,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 12130,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12577,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 13391,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13312,
            "input_tokens": 14208,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14080,
            "input_tokens": 15034,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 15862,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 16654,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 17459,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 18263,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 18414,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 19661,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20240,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 20534,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20727,
            "output_tokens": 108,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 21527,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21779,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 22696,
            "output_tokens": 120,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22950,
            "output_tokens": 258,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23901,
            "output_tokens": 121,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24717,
            "output_tokens": 1070,
            "reasoning_output_tokens": 232
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25807,
            "output_tokens": 145,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26594,
            "output_tokens": 78,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 28492,
            "output_tokens": 91,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28609,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28995,
            "output_tokens": 58,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29075,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29311,
            "output_tokens": 264,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29906,
            "output_tokens": 476,
            "reasoning_output_tokens": 208
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30659,
            "output_tokens": 399,
            "reasoning_output_tokens": 172
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31084,
            "output_tokens": 131,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31471,
            "output_tokens": 53,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31546,
            "output_tokens": 231,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32018,
            "output_tokens": 208,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32595,
            "output_tokens": 202,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33231,
            "output_tokens": 307,
            "reasoning_output_tokens": 173
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33793,
            "output_tokens": 74,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 33929,
            "output_tokens": 174,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34382,
            "output_tokens": 318,
            "reasoning_output_tokens": 171
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34956,
            "output_tokens": 146,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35292,
            "output_tokens": 254,
            "reasoning_output_tokens": 96
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35818,
            "output_tokens": 159,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36411,
            "output_tokens": 170,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36839,
            "output_tokens": 206,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37238,
            "output_tokens": 310,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37972,
            "output_tokens": 192,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38368,
            "output_tokens": 63,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38897,
            "output_tokens": 333,
            "reasoning_output_tokens": 296
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39292,
            "output_tokens": 359,
            "reasoning_output_tokens": 142
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39998,
            "output_tokens": 49,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40069,
            "output_tokens": 289,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40767,
            "output_tokens": 567,
            "reasoning_output_tokens": 325
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41360,
            "output_tokens": 150,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41768,
            "output_tokens": 257,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42373,
            "output_tokens": 432,
            "reasoning_output_tokens": 298
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43063,
            "output_tokens": 280,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43772,
            "output_tokens": 168,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43776,
            "input_tokens": 44002,
            "output_tokens": 492,
            "reasoning_output_tokens": 345
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44944,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45136,
            "output_tokens": 288,
            "reasoning_output_tokens": 155
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45835,
            "output_tokens": 396,
            "reasoning_output_tokens": 266
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46080,
            "input_tokens": 46891,
            "output_tokens": 300,
            "reasoning_output_tokens": 138
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 47848,
            "output_tokens": 348,
            "reasoning_output_tokens": 104
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48000,
            "input_tokens": 48524,
            "output_tokens": 294,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48640,
            "input_tokens": 48844,
            "output_tokens": 199,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49418,
            "output_tokens": 241,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 49990,
            "output_tokens": 44,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49920,
            "input_tokens": 50202,
            "output_tokens": 297,
            "reasoning_output_tokens": 158
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50304,
            "input_tokens": 50886,
            "output_tokens": 753,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51584,
            "input_tokens": 52015,
            "output_tokens": 330,
            "reasoning_output_tokens": 106
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52224,
            "input_tokens": 52693,
            "output_tokens": 279,
            "reasoning_output_tokens": 167
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52864,
            "input_tokens": 53161,
            "output_tokens": 339,
            "reasoning_output_tokens": 131
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 53850,
            "output_tokens": 283,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53376,
            "input_tokens": 54407,
            "output_tokens": 541,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54784,
            "input_tokens": 55119,
            "output_tokens": 266,
            "reasoning_output_tokens": 131
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55168,
            "input_tokens": 55643,
            "output_tokens": 626,
            "reasoning_output_tokens": 455
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53760,
            "input_tokens": 56564,
            "output_tokens": 185,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56064,
            "input_tokens": 56979,
            "output_tokens": 475,
            "reasoning_output_tokens": 247
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57216,
            "input_tokens": 57841,
            "output_tokens": 1257,
            "reasoning_output_tokens": 1027
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59008,
            "input_tokens": 59265,
            "output_tokens": 425,
            "reasoning_output_tokens": 212
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59648,
            "input_tokens": 60025,
            "output_tokens": 307,
            "reasoning_output_tokens": 108
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60288,
            "input_tokens": 60685,
            "output_tokens": 310,
            "reasoning_output_tokens": 176
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60928,
            "input_tokens": 61252,
            "output_tokens": 232,
            "reasoning_output_tokens": 207
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61312,
            "input_tokens": 61656,
            "output_tokens": 154,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61696,
            "input_tokens": 62067,
            "output_tokens": 219,
            "reasoning_output_tokens": 92
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62080,
            "input_tokens": 62593,
            "output_tokens": 310,
            "reasoning_output_tokens": 177
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62720,
            "input_tokens": 63157,
            "output_tokens": 244,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 63232,
            "input_tokens": 63681,
            "output_tokens": 379,
            "reasoning_output_tokens": 245
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 63872,
            "input_tokens": 64318,
            "output_tokens": 421,
            "reasoning_output_tokens": 286
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 64640,
            "input_tokens": 64996,
            "output_tokens": 1037,
            "reasoning_output_tokens": 274
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 65920,
            "input_tokens": 66057,
            "output_tokens": 92,
            "reasoning_output_tokens": 59
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 93
      },
      "analysis": {
        "observed": "Native failure after 1630 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-03/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-03/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-03/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-03/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-03/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-03/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-03/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-03/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-03/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-03/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-03/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-03/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-03/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-03/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-03/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-03/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-03/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-03/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/tabletop.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-03/workspace/.kinex/tools/tabletop.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-tabletop.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-03/workspace/.kinex/memos/robotwin-tabletop.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 189,
        "observed_images": 37
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 1700,
        "max_control_steps": 1700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 653,
        "published_frames": 653,
        "decode_ok": true,
        "duration_s": 16.325,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "665ca0a6391c6285385786c231eee6cfe15f36e59a654812c0730aabd31c5374",
        "source_sha256": {
          "third_person": "5ece68afdbd014f6964e30c61e6637f5252c3ac51c692ab0c7c177ea5aabf186",
          "left_wrist": "aee2f018aa17dd6eb4eb5304cbf193e67942b5c2be1043474aa29ca63a98aec2",
          "right_wrist": "59a9f087e664df4b9d7a90d5436c758e697d3076640e424d8e0242eaea4dc018"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "afdff04f140a",
          "task": "robotwin/put-bottles-dustbin-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 654,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 653,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 65200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 654,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 653,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 65200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 654,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 653,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 65200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-04",
      "variant": "kinex",
      "task_key": "task03/04",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "04",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 17.32,
        "step_limit": 500,
        "steps": 433,
        "success": false,
        "valid": true
      },
      "steps": 433,
      "simulation_time_s": 17.32,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 17.32,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "native_instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9682628672324576,
        "cache_read_input_tokens": 1202048,
        "cache_reported_input_tokens": 1241448,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1241448,
        "cached_input_tokens": 1202048,
        "cost_usd": null,
        "fresh_input_tokens": 39400,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1241448,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1202048,
        "known_fresh_input_tokens": 39400,
        "known_input_tokens": 1241448,
        "known_output_tokens": 10240,
        "known_reasoning_output_tokens": 4081,
        "output_tokens": 10240,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 4081,
        "reasoning_reported_output_tokens": 10240,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 44,
          "cache_write_input_tokens": 44,
          "cache_write_reported_input_tokens": 44,
          "cached_input_tokens": 44,
          "fresh_input_tokens": 44,
          "input_tokens": 44,
          "output_tokens": 44,
          "reasoning_output_tokens": 44,
          "reasoning_reported_output_tokens": 44
        },
        "request_attempts": 44,
        "response_count": 44,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1251688,
        "uncached_input_tokens": 39400,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11647,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11687,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11820,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11870,
            "output_tokens": 145,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12223,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 16733,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 17980,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18210,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 19635,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 19887,
            "output_tokens": 201,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20653,
            "output_tokens": 309,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21497,
            "output_tokens": 328,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 22472,
            "output_tokens": 383,
            "reasoning_output_tokens": 311
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 24034,
            "output_tokens": 738,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24794,
            "output_tokens": 129,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25029,
            "output_tokens": 138,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25666,
            "output_tokens": 261,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26425,
            "output_tokens": 59,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26546,
            "output_tokens": 380,
            "reasoning_output_tokens": 280
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27427,
            "output_tokens": 246,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28063,
            "output_tokens": 263,
            "reasoning_output_tokens": 175
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28374,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28956,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29538,
            "output_tokens": 154,
            "reasoning_output_tokens": 120
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29754,
            "output_tokens": 334,
            "reasoning_output_tokens": 278
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30586,
            "output_tokens": 280,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31367,
            "output_tokens": 307,
            "reasoning_output_tokens": 122
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31901,
            "output_tokens": 496,
            "reasoning_output_tokens": 400
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32465,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33140,
            "output_tokens": 78,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33716,
            "output_tokens": 204,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34416,
            "output_tokens": 246,
            "reasoning_output_tokens": 165
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 36202,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36302,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36663,
            "output_tokens": 258,
            "reasoning_output_tokens": 167
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37210,
            "output_tokens": 630,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38340,
            "output_tokens": 289,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38957,
            "output_tokens": 388,
            "reasoning_output_tokens": 275
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39843,
            "output_tokens": 308,
            "reasoning_output_tokens": 216
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40360,
            "output_tokens": 258,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40944,
            "output_tokens": 549,
            "reasoning_output_tokens": 425
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41995,
            "output_tokens": 239,
            "reasoning_output_tokens": 193
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42736,
            "output_tokens": 627,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43385,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 56
      },
      "analysis": {
        "observed": "Native failure after 433 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-04/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-04/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-04/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-04/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-04/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-04/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-04/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-04/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-04/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-04/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-04/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-04/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-04/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-04/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-04/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-04/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-04/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-04/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/dual_cartesian.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-04/workspace/.kinex/tools/dual_cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/scan-object.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-04/workspace/.kinex/memos/scan-object.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 102,
        "observed_images": 19
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 174,
        "published_frames": 174,
        "decode_ok": true,
        "duration_s": 4.35,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "7cc44b7bf3e38cad2cccf9c5e0fcb189c818d844a31a9a88e47d64ed2cd9a2c1",
        "source_sha256": {
          "third_person": "34f3844a0d2dfe8bf0ebdbd6b29e26fa05536d220c52ea0ca565e71051932d03",
          "left_wrist": "62d76135b561fcee1facb69bbd48b3456f9f1d515a63311b6ab55eba6e50c891",
          "right_wrist": "66ff60b232a2801f38ea74a7a90172a72b83951142f6cb2953ada66c6ff3a5fa"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "1ebd2ae95e55",
          "task": "robotwin/scan-object-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 175,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 174,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17320000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 175,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 174,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17320000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 175,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 174,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17320000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-05",
      "variant": "kinex",
      "task_key": "task03/05",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "05",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 41.88,
        "step_limit": 1200,
        "steps": 1047,
        "success": true,
        "valid": true
      },
      "steps": 1047,
      "simulation_time_s": 41.88,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 41.88,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "native_instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9753879435464908,
        "cache_read_input_tokens": 1202944,
        "cache_reported_input_tokens": 1233298,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1233298,
        "cached_input_tokens": 1202944,
        "cost_usd": null,
        "fresh_input_tokens": 30354,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1233298,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1202944,
        "known_fresh_input_tokens": 30354,
        "known_input_tokens": 1233298,
        "known_output_tokens": 7446,
        "known_reasoning_output_tokens": 1525,
        "output_tokens": 7446,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1525,
        "reasoning_reported_output_tokens": 7446,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 43,
          "cache_write_input_tokens": 43,
          "cache_write_reported_input_tokens": 43,
          "cached_input_tokens": 43,
          "fresh_input_tokens": 43,
          "input_tokens": 43,
          "output_tokens": 43,
          "reasoning_output_tokens": 43,
          "reasoning_reported_output_tokens": 43
        },
        "request_attempts": 43,
        "response_count": 43,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1240744,
        "uncached_input_tokens": 30354,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11636,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11676,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11809,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11857,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12105,
            "output_tokens": 282,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 18357,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 19605,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 19899,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 20090,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20669,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 20921,
            "output_tokens": 151,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 22461,
            "output_tokens": 455,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 24289,
            "output_tokens": 591,
            "reasoning_output_tokens": 181
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25516,
            "output_tokens": 160,
            "reasoning_output_tokens": 118
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 26855,
            "output_tokens": 1151,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28027,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28816,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 28926,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29554,
            "output_tokens": 141,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29722,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30098,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30685,
            "output_tokens": 66,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30778,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30969,
            "output_tokens": 99,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31316,
            "output_tokens": 252,
            "reasoning_output_tokens": 156
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31861,
            "output_tokens": 102,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32269,
            "output_tokens": 275,
            "reasoning_output_tokens": 170
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32849,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33179,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33987,
            "output_tokens": 324,
            "reasoning_output_tokens": 177
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 35049,
            "output_tokens": 98,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35519,
            "output_tokens": 145,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36013,
            "output_tokens": 453,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36691,
            "output_tokens": 252,
            "reasoning_output_tokens": 123
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37470,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37521,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37938,
            "output_tokens": 176,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38476,
            "output_tokens": 261,
            "reasoning_output_tokens": 159
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39031,
            "output_tokens": 107,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39272,
            "output_tokens": 46,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39488,
            "output_tokens": 166,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39681,
            "output_tokens": 663,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40368,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 52
      },
      "analysis": {
        "observed": "Native success after 1047 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-05/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-05/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-05/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-05/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-05/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-05/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-05/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-05/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-05/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-05/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-05/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-05/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-05/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-05/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-05/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-05/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-05/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-05/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/bowl_motion.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-05/workspace/.kinex/tools/bowl_motion.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/stack_bowls.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-05/workspace/.kinex/memos/stack_bowls.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 97,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 420,
        "published_frames": 420,
        "decode_ok": true,
        "duration_s": 10.5,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "51121401172573a5b5cf587bf69169f10282345349c548d526d10dfb8974334e",
        "source_sha256": {
          "third_person": "95a2b12ba8a89c7e5dde33535b452ba2f5b65c3be1769db02c9dbe1ccb38d8ca",
          "left_wrist": "4dfd2d049bfa38dd7339404c73b47894669d06fdb0b856cb10b91dcf6356db3a",
          "right_wrist": "1e6ea6bef59e4c42c9dc42fcd4081b1bb43f43feadf92519a21404f29f7c868b"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "6ccfc5fb9839",
          "task": "robotwin/stack-bowls-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 420,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 420,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 41880000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 420,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 420,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 41880000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 420,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 420,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 41880000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-06",
      "variant": "kinex",
      "task_key": "task03/06",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "06",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 32.56,
        "step_limit": 1200,
        "steps": 814,
        "success": true,
        "valid": true
      },
      "steps": 814,
      "simulation_time_s": 32.56,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 32.56,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "native_instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9753565460180695,
        "cache_read_input_tokens": 1666304,
        "cache_reported_input_tokens": 1708405,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1708405,
        "cached_input_tokens": 1666304,
        "cost_usd": null,
        "fresh_input_tokens": 42101,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1708405,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1666304,
        "known_fresh_input_tokens": 42101,
        "known_input_tokens": 1708405,
        "known_output_tokens": 7615,
        "known_reasoning_output_tokens": 1492,
        "output_tokens": 7615,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1492,
        "reasoning_reported_output_tokens": 7615,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 64,
          "cache_write_input_tokens": 64,
          "cache_write_reported_input_tokens": 64,
          "cached_input_tokens": 64,
          "fresh_input_tokens": 64,
          "input_tokens": 64,
          "output_tokens": 64,
          "reasoning_output_tokens": 64,
          "reasoning_reported_output_tokens": 64
        },
        "request_attempts": 64,
        "response_count": 64,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1716020,
        "uncached_input_tokens": 42101,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11645,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11693,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11824,
            "output_tokens": 21,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11859,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11895,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11953,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12104,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12551,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 13365,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13312,
            "input_tokens": 14182,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14080,
            "input_tokens": 15008,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 15836,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 16628,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 17875,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18731,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 18961,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19150,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 19729,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 19981,
            "output_tokens": 92,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20305,
            "output_tokens": 220,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20930,
            "output_tokens": 457,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21409,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 22119,
            "output_tokens": 239,
            "reasoning_output_tokens": 200
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 23053,
            "output_tokens": 151,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 24383,
            "output_tokens": 916,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25320,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25549,
            "output_tokens": 106,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 26306,
            "output_tokens": 53,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26381,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26458,
            "output_tokens": 106,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26822,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27016,
            "output_tokens": 250,
            "reasoning_output_tokens": 121
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27974,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28030,
            "output_tokens": 406,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28804,
            "output_tokens": 348,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29812,
            "output_tokens": 176,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30015,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30481,
            "output_tokens": 87,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30738,
            "output_tokens": 142,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31010,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31717,
            "output_tokens": 206,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 31951,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32663,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33026,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33219,
            "output_tokens": 264,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33889,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34596,
            "output_tokens": 65,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35432,
            "output_tokens": 109,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35673,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35714,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35835,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36216,
            "output_tokens": 155,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36621,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36812,
            "output_tokens": 40,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37021,
            "output_tokens": 212,
            "reasoning_output_tokens": 112
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37973,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38031,
            "output_tokens": 109,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38345,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38537,
            "output_tokens": 373,
            "reasoning_output_tokens": 231
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39044,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39137,
            "output_tokens": 50,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39357,
            "output_tokens": 168,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 39552,
            "output_tokens": 584,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40159,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 63
      },
      "analysis": {
        "observed": "Native success after 814 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-06/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-06/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-06/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-06/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-06/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-06/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-06/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-06/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-06/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-06/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-06/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-06/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-06/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-06/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-06/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-06/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-06/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-06/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-06/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/stack_perception.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-06/workspace/.kinex/tools/stack_perception.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/aloha_stack.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-06/workspace/.kinex/memos/aloha_stack.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 129,
        "observed_images": 13
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 327,
        "published_frames": 327,
        "decode_ok": true,
        "duration_s": 8.175,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "9fb569dcc5dd93074137e691fd22b650fb04d209af505b114f7a457cdc26aa48",
        "source_sha256": {
          "third_person": "a3929607d5b72bebdd7240719c1f4dcec493c77f291db33c2adca6502e4cb364",
          "left_wrist": "e1f2d2c23ced434e9032cc036e9d386649b2886c3f6c2055482172e8b3e7313a",
          "right_wrist": "af4e0b4bee2491829f610f869095fede92d43324ce2a4d9e473b987b6e2f5de0"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "e6fcedfec37b",
          "task": "robotwin/stack-blocks-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 327,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 327,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 32560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 327,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 327,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 32560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 327,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 327,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 32560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-07",
      "variant": "kinex",
      "task_key": "task03/07",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "07",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 22.36,
        "step_limit": 900,
        "steps": 559,
        "success": false,
        "valid": true
      },
      "steps": 559,
      "simulation_time_s": 22.36,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 22.36,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the cylindrical mug with handle on the table, rotate it, set it down in the middle, then hang it onto the rack with straight and slanted rods.",
      "native_instruction": "Grab the cylindrical mug with handle on the table, rotate it, set it down in the middle, then hang it onto the rack with straight and slanted rods.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9689616248154841,
        "cache_read_input_tokens": 1171712,
        "cache_reported_input_tokens": 1209245,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1209245,
        "cached_input_tokens": 1171712,
        "cost_usd": null,
        "fresh_input_tokens": 37533,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1209245,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1171712,
        "known_fresh_input_tokens": 37533,
        "known_input_tokens": 1209245,
        "known_output_tokens": 7708,
        "known_reasoning_output_tokens": 2434,
        "output_tokens": 7708,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2434,
        "reasoning_reported_output_tokens": 7708,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 43,
          "cache_write_input_tokens": 43,
          "cache_write_reported_input_tokens": 43,
          "cached_input_tokens": 43,
          "fresh_input_tokens": 43,
          "input_tokens": 43,
          "output_tokens": 43,
          "reasoning_output_tokens": 43,
          "reasoning_reported_output_tokens": 43
        },
        "request_attempts": 43,
        "response_count": 43,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1216953,
        "uncached_input_tokens": 37533,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11654,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11787,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11835,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 12083,
            "output_tokens": 192,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 15918,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 17165,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17024,
            "input_tokens": 17744,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 17974,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18166,
            "output_tokens": 154,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 20291,
            "output_tokens": 387,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20096,
            "input_tokens": 22187,
            "output_tokens": 466,
            "reasoning_output_tokens": 426
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 23119,
            "output_tokens": 916,
            "reasoning_output_tokens": 116
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 24056,
            "output_tokens": 125,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24804,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 24883,
            "output_tokens": 125,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25382,
            "output_tokens": 66,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25820,
            "output_tokens": 131,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26323,
            "output_tokens": 231,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 27244,
            "output_tokens": 184,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27817,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28229,
            "output_tokens": 199,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28955,
            "output_tokens": 171,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29790,
            "output_tokens": 142,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30859,
            "output_tokens": 235,
            "reasoning_output_tokens": 197
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31789,
            "output_tokens": 485,
            "reasoning_output_tokens": 349
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32651,
            "output_tokens": 55,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32728,
            "output_tokens": 243,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33479,
            "output_tokens": 203,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33709,
            "output_tokens": 143,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34228,
            "output_tokens": 109,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34715,
            "output_tokens": 74,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35159,
            "output_tokens": 49,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35270,
            "output_tokens": 87,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35727,
            "output_tokens": 87,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36300,
            "output_tokens": 100,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 37171,
            "output_tokens": 159,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37348,
            "output_tokens": 132,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37856,
            "output_tokens": 243,
            "reasoning_output_tokens": 138
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38473,
            "output_tokens": 318,
            "reasoning_output_tokens": 213
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 39166,
            "output_tokens": 174,
            "reasoning_output_tokens": 133
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39402,
            "output_tokens": 209,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39638,
            "output_tokens": 690,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 40351,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 50
      },
      "analysis": {
        "observed": "Native failure after 559 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-07/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-07/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-07/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-07/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-07/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-07/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-07/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-07/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-07/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-07/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-07/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-07/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-07/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-07/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-07/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-07/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-07/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-07/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/mug_motion.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-07/workspace/.kinex/tools/mug_motion.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/hanging-mug.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-07/workspace/.kinex/memos/hanging-mug.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 95,
        "observed_images": 17
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 225,
        "published_frames": 225,
        "decode_ok": true,
        "duration_s": 5.625,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "21070409dee278838101c9918d3953e4f940d4d7515bb40b085d5406beae541c",
        "source_sha256": {
          "third_person": "41d7d3fc602d74b11bc33e88118b4ad8472fc1212a880489491e18094de20751",
          "left_wrist": "f5e6aaab832a669f010dc93a2fdf25137c9ecf28c0f33cccf3dda5dcf77cecd1",
          "right_wrist": "5b7c6afc7d2eb4033bd48ae84383d38ac5b6fda4778057004071b0642b5cb8e6"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "79d65268e0c9",
          "task": "robotwin/hanging-mug-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 225,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 225,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 22360000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 225,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 225,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 22360000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 225,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 225,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 22360000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-08",
      "variant": "kinex",
      "task_key": "task03/08",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "08",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "step_limit",
        "simulation_time_s": 28.0,
        "step_limit": 700,
        "steps": 700,
        "success": false,
        "valid": true
      },
      "steps": 700,
      "simulation_time_s": 28.0,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 28.0,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Open the cabinet with smooth finish's drawer and set the blue soap with printed patterns inside with the other arm.",
      "native_instruction": "Open the cabinet with smooth finish's drawer and set the blue soap with printed patterns inside with the other arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9777702654884282,
        "cache_read_input_tokens": 2846080,
        "cache_reported_input_tokens": 2910786,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2910786,
        "cached_input_tokens": 2846080,
        "cost_usd": null,
        "fresh_input_tokens": 64706,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2910786,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2846080,
        "known_fresh_input_tokens": 64706,
        "known_input_tokens": 2910786,
        "known_output_tokens": 16258,
        "known_reasoning_output_tokens": 7689,
        "output_tokens": 16258,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 7689,
        "reasoning_reported_output_tokens": 16258,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 65,
          "cache_write_input_tokens": 65,
          "cache_write_reported_input_tokens": 65,
          "cached_input_tokens": 65,
          "fresh_input_tokens": 65,
          "input_tokens": 65,
          "output_tokens": 65,
          "reasoning_output_tokens": 65,
          "reasoning_reported_output_tokens": 65
        },
        "request_attempts": 65,
        "response_count": 65,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2927044,
        "uncached_input_tokens": 64706,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11644,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11684,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11817,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 15762,
            "output_tokens": 264,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 19702,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20949,
            "output_tokens": 162,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 22593,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22784,
            "output_tokens": 210,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 25206,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 27103,
            "output_tokens": 57,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27223,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27580,
            "output_tokens": 320,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28295,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28566,
            "output_tokens": 262,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29292,
            "output_tokens": 265,
            "reasoning_output_tokens": 223
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30736,
            "output_tokens": 1181,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31939,
            "output_tokens": 250,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32954,
            "output_tokens": 61,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33077,
            "output_tokens": 73,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33845,
            "output_tokens": 656,
            "reasoning_output_tokens": 199
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34990,
            "output_tokens": 212,
            "reasoning_output_tokens": 142
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35230,
            "output_tokens": 146,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 36101,
            "output_tokens": 167,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 37042,
            "output_tokens": 266,
            "reasoning_output_tokens": 241
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37478,
            "output_tokens": 208,
            "reasoning_output_tokens": 127
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38259,
            "output_tokens": 278,
            "reasoning_output_tokens": 163
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 39434,
            "output_tokens": 504,
            "reasoning_output_tokens": 382
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40665,
            "output_tokens": 177,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41571,
            "output_tokens": 102,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 43505,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 43661,
            "output_tokens": 120,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44752,
            "output_tokens": 138,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45661,
            "output_tokens": 334,
            "reasoning_output_tokens": 210
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45952,
            "input_tokens": 46764,
            "output_tokens": 285,
            "reasoning_output_tokens": 184
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47732,
            "output_tokens": 151,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 48573,
            "output_tokens": 597,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49224,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49152,
            "input_tokens": 49997,
            "output_tokens": 248,
            "reasoning_output_tokens": 164
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50176,
            "input_tokens": 50751,
            "output_tokens": 117,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50816,
            "input_tokens": 51436,
            "output_tokens": 400,
            "reasoning_output_tokens": 318
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51712,
            "input_tokens": 52411,
            "output_tokens": 231,
            "reasoning_output_tokens": 145
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52480,
            "input_tokens": 53220,
            "output_tokens": 369,
            "reasoning_output_tokens": 278
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53504,
            "input_tokens": 54164,
            "output_tokens": 152,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54144,
            "input_tokens": 55281,
            "output_tokens": 287,
            "reasoning_output_tokens": 152
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55424,
            "input_tokens": 56337,
            "output_tokens": 262,
            "reasoning_output_tokens": 181
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56448,
            "input_tokens": 57163,
            "output_tokens": 535,
            "reasoning_output_tokens": 454
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57472,
            "input_tokens": 58266,
            "output_tokens": 258,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58496,
            "input_tokens": 59290,
            "output_tokens": 621,
            "reasoning_output_tokens": 353
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59776,
            "input_tokens": 60361,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60288,
            "input_tokens": 60696,
            "output_tokens": 282,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60800,
            "input_tokens": 61172,
            "output_tokens": 294,
            "reasoning_output_tokens": 217
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61312,
            "input_tokens": 61529,
            "output_tokens": 117,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61440,
            "input_tokens": 62150,
            "output_tokens": 116,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62080,
            "input_tokens": 62918,
            "output_tokens": 270,
            "reasoning_output_tokens": 144
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62976,
            "input_tokens": 63842,
            "output_tokens": 99,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 63744,
            "input_tokens": 64314,
            "output_tokens": 122,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 64256,
            "input_tokens": 64458,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 64384,
            "input_tokens": 65011,
            "output_tokens": 295,
            "reasoning_output_tokens": 166
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 65280,
            "input_tokens": 66069,
            "output_tokens": 607,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 66560,
            "input_tokens": 67182,
            "output_tokens": 148,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 67200,
            "input_tokens": 67836,
            "output_tokens": 485,
            "reasoning_output_tokens": 390
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 68224,
            "input_tokens": 68894,
            "output_tokens": 578,
            "reasoning_output_tokens": 487
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 69376,
            "input_tokens": 70042,
            "output_tokens": 135,
            "reasoning_output_tokens": 75
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 70016,
            "input_tokens": 70775,
            "output_tokens": 1026,
            "reasoning_output_tokens": 196
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 71680,
            "input_tokens": 71828,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 77
      },
      "analysis": {
        "observed": "Native failure after 700 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-08/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-08/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-08/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-08/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-08/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-08/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-08/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-08/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-08/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-08/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-08/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-08/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-08/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-08/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-08/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-08/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-08/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-08/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-08/workspace/.kinex/tools/cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/cabinet-manipulation.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-08/workspace/.kinex/memos/cabinet-manipulation.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 144,
        "observed_images": 40
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 281,
        "published_frames": 281,
        "decode_ok": true,
        "duration_s": 7.025,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f0d05e1b7ca212c1e21137f559e208f91c9aff4e5b369c48ab342cd558d3aac6",
        "source_sha256": {
          "third_person": "03b7194afa397823c5ff59c63ee1e077c48605fc5812db911bb0c62312a36a10",
          "left_wrist": "437435231063fbb8e3a3c5944dd53a2680ecfdf7dd99c44ec785882873dc64e1",
          "right_wrist": "4d02cdccbf80c320c5bd6c103e73b8cde23f1fa8236abb478c1ee733046c57a1"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "15ea5fe013be",
          "task": "robotwin/put-object-cabinet-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 282,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 281,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28000000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 282,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 281,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28000000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 282,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 281,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 28000000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-09",
      "variant": "kinex",
      "task_key": "task03/09",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "09",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 30.240000000000002,
        "step_limit": 1200,
        "steps": 756,
        "success": true,
        "valid": true
      },
      "steps": 756,
      "simulation_time_s": 30.240000000000002,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 30.240000000000002,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move large block, medium block, and small block to the center table.",
      "native_instruction": "Move large block, medium block, and small block to the center table.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9690485062196635,
        "cache_read_input_tokens": 785408,
        "cache_reported_input_tokens": 810494,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 810494,
        "cached_input_tokens": 785408,
        "cost_usd": null,
        "fresh_input_tokens": 25086,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 810494,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 785408,
        "known_fresh_input_tokens": 25086,
        "known_input_tokens": 810494,
        "known_output_tokens": 5480,
        "known_reasoning_output_tokens": 870,
        "output_tokens": 5480,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 870,
        "reasoning_reported_output_tokens": 5480,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 31,
          "cache_write_input_tokens": 31,
          "cache_write_reported_input_tokens": 31,
          "cached_input_tokens": 31,
          "fresh_input_tokens": 31,
          "input_tokens": 31,
          "output_tokens": 31,
          "reasoning_output_tokens": 31,
          "reasoning_reported_output_tokens": 31
        },
        "request_attempts": 31,
        "response_count": 31,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 815974,
        "uncached_input_tokens": 25086,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11636,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11676,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11809,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 14610,
            "output_tokens": 199,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14720,
            "input_tokens": 17691,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 18939,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19169,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 21535,
            "output_tokens": 118,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 21863,
            "output_tokens": 137,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22774,
            "output_tokens": 290,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23599,
            "output_tokens": 305,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24522,
            "output_tokens": 190,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 26816,
            "output_tokens": 1212,
            "reasoning_output_tokens": 279
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28051,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28650,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28703,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28787,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29012,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29151,
            "output_tokens": 180,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29752,
            "output_tokens": 163,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30444,
            "output_tokens": 176,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30914,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31585,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32230,
            "output_tokens": 149,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32902,
            "output_tokens": 157,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33197,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33289,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33483,
            "output_tokens": 380,
            "reasoning_output_tokens": 207
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34123,
            "output_tokens": 369,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34521,
            "output_tokens": 517,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35061,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 41
      },
      "analysis": {
        "observed": "Native success after 756 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-09/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-09/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-09/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-09/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-09/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-09/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-09/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-09/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-09/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-09/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-09/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-09/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-09/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-09/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-09/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-09/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-09/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-09/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/aloha_cartesian.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-09/workspace/.kinex/tools/aloha_cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/aloha_blocks.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-09/workspace/.kinex/memos/aloha_blocks.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 74,
        "observed_images": 9
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 304,
        "published_frames": 304,
        "decode_ok": true,
        "duration_s": 7.6,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "1def608cc03ed8695834f122f41fa568ea3d655379a9534546b858b8f9fa12e7",
        "source_sha256": {
          "third_person": "422a0e829afb2ff4dc267999f64a90738c04812728724d4540c9132f40e7032c",
          "left_wrist": "bb132ee8e17fc21d1d9f179b78b4692818ded7bda6def8e49646f142e0a5975f",
          "right_wrist": "3b9c0200caf8728b52e92fba29f5665b47caf8b6652bc210d5ebb67b232eab92"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "62b4ee944356",
          "task": "robotwin/blocks-ranking-size-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 304,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 304,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 30240000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 304,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 304,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 30240000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 304,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 304,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 30240000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-10",
      "variant": "kinex",
      "task_key": "task03/10",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "10",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 7.84,
        "step_limit": 400,
        "steps": 196,
        "success": true,
        "valid": true
      },
      "steps": 196,
      "simulation_time_s": 7.84,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 7.84,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Hold the dark gray cooking kitchenpot firmly and lift it",
      "native_instruction": "Hold the dark gray cooking kitchenpot firmly and lift it",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.914501358150229,
        "cache_read_input_tokens": 585472,
        "cache_reported_input_tokens": 640209,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 640209,
        "cached_input_tokens": 585472,
        "cost_usd": null,
        "fresh_input_tokens": 54737,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 640209,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 585472,
        "known_fresh_input_tokens": 54737,
        "known_input_tokens": 640209,
        "known_output_tokens": 5149,
        "known_reasoning_output_tokens": 720,
        "output_tokens": 5149,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 720,
        "reasoning_reported_output_tokens": 5149,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 28,
          "cache_write_input_tokens": 28,
          "cache_write_reported_input_tokens": 28,
          "cached_input_tokens": 28,
          "fresh_input_tokens": 28,
          "input_tokens": 28,
          "output_tokens": 28,
          "reasoning_output_tokens": 28,
          "reasoning_reported_output_tokens": 28
        },
        "request_attempts": 28,
        "response_count": 28,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 645358,
        "uncached_input_tokens": 54737,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11632,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11680,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11811,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12059,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12873,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12928,
            "input_tokens": 16720,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 17968,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18198,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18390,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 18969,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19221,
            "output_tokens": 380,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 19996,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 20267,
            "output_tokens": 368,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 21974,
            "output_tokens": 268,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 24091,
            "output_tokens": 128,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24685,
            "output_tokens": 729,
            "reasoning_output_tokens": 251
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25786,
            "output_tokens": 74,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26555,
            "output_tokens": 964,
            "reasoning_output_tokens": 135
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27541,
            "output_tokens": 120,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27963,
            "output_tokens": 132,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28956,
            "output_tokens": 185,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 30018,
            "output_tokens": 129,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 31019,
            "output_tokens": 65,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31731,
            "output_tokens": 160,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 32025,
            "output_tokens": 62,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32149,
            "output_tokens": 165,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32663,
            "output_tokens": 584,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33269,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 32
      },
      "analysis": {
        "observed": "Native success after 196 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-10/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-10/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-10/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-10/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-10/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-10/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-10/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-10/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-10/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-10/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-10/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-10/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-10/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-10/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-10/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-10/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-10/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-10/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/dual_pot.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-10/workspace/.kinex/tools/dual_pot.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/lift-pot.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-10/workspace/.kinex/memos/lift-pot.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 62,
        "observed_images": 7
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 80,
        "published_frames": 80,
        "decode_ok": true,
        "duration_s": 2.0,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "2477080981c17f33ea46b8e272cfbe09732b6eb9199f1a78b300b32cfcd24f81",
        "source_sha256": {
          "third_person": "0daa6bccc2bd290838ed91c57bc84f91ccd2c13aa9ea5e6e0577de4d383d8248",
          "left_wrist": "f5c48bf8c7b0d7643f2c7e6eb5aa332c6142c8aada1a5cde9c9d9bbdf7afa8d7",
          "right_wrist": "1c611b8a4ef4d16bc6f629b23cc90cb87ac6ad953f2ce68060a2935d799421ef"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "0229c575b940",
          "task": "robotwin/lift-pot-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 80,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 80,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7840000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 80,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 80,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7840000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 80,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 80,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7840000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-11",
      "variant": "kinex",
      "task_key": "task03/11",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "11",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 2
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 9.08,
        "step_limit": 400,
        "steps": 227,
        "success": false,
        "valid": true
      },
      "steps": 227,
      "simulation_time_s": 9.08,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 9.08,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the green plastic bottle with ridged bottom carefully using the right arm",
      "native_instruction": "Pick up the green plastic bottle with ridged bottom carefully using the right arm",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.926035612067376,
        "cache_read_input_tokens": 577536,
        "cache_reported_input_tokens": 623665,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 623665,
        "cached_input_tokens": 577536,
        "cost_usd": null,
        "fresh_input_tokens": 46129,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 623665,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 577536,
        "known_fresh_input_tokens": 46129,
        "known_input_tokens": 623665,
        "known_output_tokens": 4905,
        "known_reasoning_output_tokens": 588,
        "output_tokens": 4905,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 588,
        "reasoning_reported_output_tokens": 4905,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 30,
          "cache_write_input_tokens": 30,
          "cache_write_reported_input_tokens": 30,
          "cached_input_tokens": 30,
          "fresh_input_tokens": 30,
          "input_tokens": 30,
          "output_tokens": 30,
          "reasoning_output_tokens": 30,
          "reasoning_reported_output_tokens": 30
        },
        "request_attempts": 30,
        "response_count": 30,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 628570,
        "uncached_input_tokens": 46129,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11633,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11766,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11814,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 12062,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12879,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 15095,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14976,
            "input_tokens": 16342,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16256,
            "input_tokens": 16506,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 16696,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 17275,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17152,
            "input_tokens": 17527,
            "output_tokens": 118,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 18804,
            "output_tokens": 147,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19653,
            "output_tokens": 520,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20705,
            "output_tokens": 658,
            "reasoning_output_tokens": 124
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21384,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21601,
            "output_tokens": 454,
            "reasoning_output_tokens": 173
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22598,
            "output_tokens": 50,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 23715,
            "output_tokens": 284,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24287,
            "output_tokens": 808,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25116,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25395,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25588,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25868,
            "output_tokens": 88,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26007,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26386,
            "output_tokens": 176,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26744,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27028,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27220,
            "output_tokens": 237,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12928,
            "input_tokens": 27755,
            "output_tokens": 437,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28216,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 33
      },
      "analysis": {
        "observed": "Native failure after 227 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-11/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-11/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-11/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-11/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-11/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-11/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-11/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-11/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-11/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-11/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-11/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-11/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-11/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-11/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-11/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-11/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-11/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-11/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/bottle_geometry.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-11/workspace/.kinex/tools/bottle_geometry.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/right_tcp_motion.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-11/workspace/.kinex/tools/right_tcp_motion.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/aloha_bottle.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-11/workspace/.kinex/memos/aloha_bottle.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 65,
        "observed_images": 5
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 92,
        "published_frames": 92,
        "decode_ok": true,
        "duration_s": 2.3,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "2256f8b97a435fbf03a5f16c72dd452e7c2cfed8e9fcdf67214b3b847ac62773",
        "source_sha256": {
          "third_person": "c0217142cf8b63bebaaa5a313e49af3acfb7f40e6c4fd6f326e676e4a3f6bec3",
          "left_wrist": "dd71b36337d1f046ba6c3e92b0e1e5a6aa54942951614d3b71eb0dc877840c1d",
          "right_wrist": "15cc772335ac7cc1f37ea5bda93e0b35f6ffbfa6fa6cd8164a1f38bf6877bf57"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "3f8defed1e2f",
          "task": "robotwin/adjust-bottle-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 92,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 92,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9080000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 92,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 92,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9080000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 92,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 92,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9080000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-12",
      "variant": "kinex",
      "task_key": "task03/12",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "12",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.44,
        "step_limit": 400,
        "steps": 236,
        "success": true,
        "valid": true
      },
      "steps": 236,
      "simulation_time_s": 9.44,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 9.44,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the hammer with black handle and hit the block",
      "native_instruction": "Grab the hammer with black handle and hit the block",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9607224975611556,
        "cache_read_input_tokens": 833152,
        "cache_reported_input_tokens": 867214,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 867214,
        "cached_input_tokens": 833152,
        "cost_usd": null,
        "fresh_input_tokens": 34062,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 867214,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 833152,
        "known_fresh_input_tokens": 34062,
        "known_input_tokens": 867214,
        "known_output_tokens": 5102,
        "known_reasoning_output_tokens": 936,
        "output_tokens": 5102,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 936,
        "reasoning_reported_output_tokens": 5102,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 35,
          "cache_write_input_tokens": 35,
          "cache_write_reported_input_tokens": 35,
          "cached_input_tokens": 35,
          "fresh_input_tokens": 35,
          "input_tokens": 35,
          "output_tokens": 35,
          "reasoning_output_tokens": 35,
          "reasoning_reported_output_tokens": 35
        },
        "request_attempts": 35,
        "response_count": 35,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 872316,
        "uncached_input_tokens": 34062,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11633,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11673,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11806,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11854,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 13831,
            "output_tokens": 192,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13824,
            "input_tokens": 17666,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 18913,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19109,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 19696,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20275,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20096,
            "input_tokens": 20527,
            "output_tokens": 351,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21273,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21544,
            "output_tokens": 277,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 22790,
            "output_tokens": 148,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 24886,
            "output_tokens": 103,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25007,
            "output_tokens": 222,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25867,
            "output_tokens": 293,
            "reasoning_output_tokens": 126
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26493,
            "output_tokens": 167,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26790,
            "output_tokens": 820,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27631,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28073,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28509,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28911,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29681,
            "output_tokens": 60,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29763,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30676,
            "output_tokens": 108,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30811,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31253,
            "output_tokens": 214,
            "reasoning_output_tokens": 172
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31581,
            "output_tokens": 157,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32088,
            "output_tokens": 167,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32615,
            "output_tokens": 341,
            "reasoning_output_tokens": 255
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33090,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33147,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33595,
            "output_tokens": 541,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34157,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 41
      },
      "analysis": {
        "observed": "Native success after 236 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-12/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-12/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-12/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-12/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-12/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-12/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-12/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-12/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-12/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-12/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-12/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-12/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-12/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-12/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-12/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-12/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-12/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-12/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-12/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/hammer.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-12/workspace/.kinex/memos/hammer.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 78,
        "observed_images": 9
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 96,
        "published_frames": 96,
        "decode_ok": true,
        "duration_s": 2.4,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "76f8d1ae9c74928483343ce192292d43bbbc10b84680baa01050dfd03b8a7174",
        "source_sha256": {
          "third_person": "bceb081bfc1dbab97c266371587fe5ed2ffe1fb9cee624e309677f5ab0c88737",
          "left_wrist": "23c40751df9e67df45f3d690e7997de98942ff5d6b2b8066d2e1b5d691c90115",
          "right_wrist": "245fcba45b245b03518344fbb505050bf4e3047bfa203d60ff5a866e248c8ad3"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "89b8129985a8",
          "task": "robotwin/beat-block-hammer-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 96,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 96,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9440000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 96,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 96,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9440000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 96,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 96,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9440000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-13",
      "variant": "kinex",
      "task_key": "task03/13",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "13",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 32.2,
        "step_limit": 1200,
        "steps": 805,
        "success": true,
        "valid": true
      },
      "steps": 805,
      "simulation_time_s": 32.2,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 32.2,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set red block on the left, then green block in the center, and finally blue block on the right.",
      "native_instruction": "Set red block on the left, then green block in the center, and finally blue block on the right.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9569688224268396,
        "cache_read_input_tokens": 1446400,
        "cache_reported_input_tokens": 1511439,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1511439,
        "cached_input_tokens": 1446400,
        "cost_usd": null,
        "fresh_input_tokens": 65039,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1511439,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1446400,
        "known_fresh_input_tokens": 65039,
        "known_input_tokens": 1511439,
        "known_output_tokens": 6563,
        "known_reasoning_output_tokens": 1035,
        "output_tokens": 6563,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1035,
        "reasoning_reported_output_tokens": 6563,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 48,
          "cache_write_input_tokens": 48,
          "cache_write_reported_input_tokens": 48,
          "cached_input_tokens": 48,
          "fresh_input_tokens": 48,
          "input_tokens": 48,
          "output_tokens": 48,
          "reasoning_output_tokens": 48,
          "reasoning_reported_output_tokens": 48
        },
        "request_attempts": 48,
        "response_count": 48,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1518002,
        "uncached_input_tokens": 65039,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11645,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11685,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11818,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 15810,
            "output_tokens": 246,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 19732,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20979,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21209,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 24360,
            "output_tokens": 214,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25451,
            "output_tokens": 378,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 26798,
            "output_tokens": 200,
            "reasoning_output_tokens": 160
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27464,
            "output_tokens": 1519,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29005,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29564,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30264,
            "output_tokens": 103,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30395,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30723,
            "output_tokens": 97,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31083,
            "output_tokens": 43,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31296,
            "output_tokens": 87,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31488,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31826,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32019,
            "output_tokens": 138,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32409,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32566,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 32915,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33473,
            "output_tokens": 89,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33823,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 33981,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34316,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34509,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34858,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35015,
            "output_tokens": 240,
            "reasoning_output_tokens": 144
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35511,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36069,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36580,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36919,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37253,
            "output_tokens": 144,
            "reasoning_output_tokens": 104
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37419,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37577,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37911,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38103,
            "output_tokens": 105,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38466,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38660,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38996,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39182,
            "output_tokens": 88,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39332,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39784,
            "output_tokens": 468,
            "reasoning_output_tokens": 221
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40280,
            "output_tokens": 616,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 40918,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 61
      },
      "analysis": {
        "observed": "Native success after 805 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-13/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-13/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-13/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-13/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-13/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-13/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-13/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-13/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-13/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-13/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-13/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-13/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-13/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-13/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-13/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-13/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-13/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-13/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/block_manipulation.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-13/workspace/.kinex/tools/block_manipulation.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/blocks-ranking.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-13/workspace/.kinex/memos/blocks-ranking.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 111,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 323,
        "published_frames": 323,
        "decode_ok": true,
        "duration_s": 8.075,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "e268f69e0dc1a47255585de26efebe09c85eae4b25d7d5458b6195244cf87716",
        "source_sha256": {
          "third_person": "14909d94b802182394d47cfbbe5a0af8d1b36cfb6cd6040b8f8eaad0f9df8889",
          "left_wrist": "39bc4e3222e6c86f2616e8c4ee87bc8d086b4a929890e741012ff1d9d7533511",
          "right_wrist": "42d967c9bdb07ead3fdd68c943056d172009d91a4a4d08d40a12d0caa2d76f15"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "0b1e1007af88",
          "task": "robotwin/blocks-ranking-rgb-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 324,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 323,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 32200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 324,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 323,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 32200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 324,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 323,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 32200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-14",
      "variant": "kinex",
      "task_key": "task03/14",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "14",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 4.5600000000000005,
        "step_limit": 400,
        "steps": 114,
        "success": true,
        "valid": true
      },
      "steps": 114,
      "simulation_time_s": 4.5600000000000005,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 4.5600000000000005,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Click the top button of the alarm-clock with beige screen",
      "native_instruction": "Click the top button of the alarm-clock with beige screen",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9017010164790582,
        "cache_read_input_tokens": 611200,
        "cache_reported_input_tokens": 677830,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 677830,
        "cached_input_tokens": 611200,
        "cost_usd": null,
        "fresh_input_tokens": 66630,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 677830,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 611200,
        "known_fresh_input_tokens": 66630,
        "known_input_tokens": 677830,
        "known_output_tokens": 4192,
        "known_reasoning_output_tokens": 695,
        "output_tokens": 4192,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 695,
        "reasoning_reported_output_tokens": 4192,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 27,
          "cache_write_input_tokens": 27,
          "cache_write_reported_input_tokens": 27,
          "cached_input_tokens": 27,
          "fresh_input_tokens": 27,
          "input_tokens": 27,
          "output_tokens": 27,
          "reasoning_output_tokens": 27,
          "reasoning_reported_output_tokens": 27
        },
        "request_attempts": 27,
        "response_count": 27,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 682022,
        "uncached_input_tokens": 66630,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11632,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11672,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11805,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 15108,
            "output_tokens": 186,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15104,
            "input_tokens": 17399,
            "output_tokens": 145,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 20447,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20677,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 20868,
            "output_tokens": 80,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21725,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 21977,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22248,
            "output_tokens": 218,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23001,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23876,
            "output_tokens": 148,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 26016,
            "output_tokens": 53,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26132,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26489,
            "output_tokens": 196,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 29021,
            "output_tokens": 324,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30177,
            "output_tokens": 1297,
            "reasoning_output_tokens": 318
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31494,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 31858,
            "output_tokens": 126,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32581,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32663,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33113,
            "output_tokens": 240,
            "reasoning_output_tokens": 154
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33492,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33549,
            "output_tokens": 75,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 34132,
            "output_tokens": 520,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34678,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 40
      },
      "analysis": {
        "observed": "Native success after 114 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-14/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-14/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-14/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-14/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-14/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-14/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-14/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-14/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-14/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-14/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-14/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-14/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-14/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-14/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-14/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-14/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-14/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-14/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-14/workspace/.kinex/tools/cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-click-alarmclock.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-14/workspace/.kinex/memos/robotwin-click-alarmclock.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 69,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 47,
        "published_frames": 47,
        "decode_ok": true,
        "duration_s": 1.175,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "3168af243772bc078792921a5d336b23683c0e2b4ff28c4ae343df2431d89c97",
        "source_sha256": {
          "third_person": "b0a8309a2f7aa95618e376f330ee3824b293d387fa82d79b37340d89c57f9d7e",
          "left_wrist": "8cbf759e3fcb3780e1573648b5aeeb5e8d4bd0809e29150c415dc6217275555c",
          "right_wrist": "b1f02201b9a81f249d545f2674dcb9b0f610076fa037b79fa980fec69ab4b832"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "0f4567cdb4d4",
          "task": "robotwin/click-alarmclock-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 47,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 47,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 4560000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 47,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 47,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 4560000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 47,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 47,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 4560000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-15",
      "variant": "kinex",
      "task_key": "task03/15",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "15",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 10.84,
        "step_limit": 400,
        "steps": 271,
        "success": false,
        "valid": true
      },
      "steps": 271,
      "simulation_time_s": 10.84,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 10.84,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to press the medium-sized bell's top center.",
      "native_instruction": "Use the left arm to press the medium-sized bell's top center.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9498824841444794,
        "cache_read_input_tokens": 546816,
        "cache_reported_input_tokens": 575667,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 575667,
        "cached_input_tokens": 546816,
        "cost_usd": null,
        "fresh_input_tokens": 28851,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 575667,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 546816,
        "known_fresh_input_tokens": 28851,
        "known_input_tokens": 575667,
        "known_output_tokens": 5330,
        "known_reasoning_output_tokens": 1099,
        "output_tokens": 5330,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1099,
        "reasoning_reported_output_tokens": 5330,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 27,
          "cache_write_input_tokens": 27,
          "cache_write_reported_input_tokens": 27,
          "cached_input_tokens": 27,
          "fresh_input_tokens": 27,
          "input_tokens": 27,
          "output_tokens": 27,
          "reasoning_output_tokens": 27,
          "reasoning_reported_output_tokens": 27
        },
        "request_attempts": 27,
        "response_count": 27,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 580997,
        "uncached_input_tokens": 28851,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11633,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11681,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11812,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12060,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12507,
            "output_tokens": 166,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 15887,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 17134,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17024,
            "input_tokens": 17298,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17152,
            "input_tokens": 17755,
            "output_tokens": 77,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 18372,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 18624,
            "output_tokens": 283,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 19302,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 19573,
            "output_tokens": 284,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 21010,
            "output_tokens": 254,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 22011,
            "output_tokens": 145,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 23315,
            "output_tokens": 384,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24104,
            "output_tokens": 436,
            "reasoning_output_tokens": 124
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25035,
            "output_tokens": 818,
            "reasoning_output_tokens": 166
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25875,
            "output_tokens": 134,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26466,
            "output_tokens": 404,
            "reasoning_output_tokens": 127
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27278,
            "output_tokens": 212,
            "reasoning_output_tokens": 104
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27955,
            "output_tokens": 174,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28601,
            "output_tokens": 170,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29236,
            "output_tokens": 260,
            "reasoning_output_tokens": 152
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29967,
            "output_tokens": 209,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30316,
            "output_tokens": 521,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30860,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 31
      },
      "analysis": {
        "observed": "Native failure after 271 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-15/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-15/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-15/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-15/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-15/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-15/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-15/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-15/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-15/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-15/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-15/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-15/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-15/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-15/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-15/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-15/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-15/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-15/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/left_cartesian.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-15/workspace/.kinex/tools/left_cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/click_bell.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-15/workspace/.kinex/memos/click_bell.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 60,
        "observed_images": 7
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 110,
        "published_frames": 110,
        "decode_ok": true,
        "duration_s": 2.75,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f1165ba32395d0cc2bad178bd8df2f1ef845f6c4665c05ccab97d85e1828bc6e",
        "source_sha256": {
          "third_person": "a0c7488c935c22583cb69430e60f5efc574aaef93195d2288c57ac22fc992d62",
          "left_wrist": "bb3c8f527b7634bd95464131b5f1cb189a1018be5a4900a18c0fa7ab1aa806b8",
          "right_wrist": "b7b8a811834be20dcff13533f317789245e764984744e42389a4190daeb20e04"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "bba71ea01f46",
          "task": "robotwin/click-bell-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 110,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 110,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10840000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 110,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 110,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10840000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 110,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 110,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10840000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-16",
      "variant": "kinex",
      "task_key": "task03/16",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "16",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 19.48,
        "step_limit": 600,
        "steps": 487,
        "success": false,
        "valid": true
      },
      "steps": 487,
      "simulation_time_s": 19.48,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 19.48,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the compact tabletop trashbin, pour balls into the bin.",
      "native_instruction": "Pick up the compact tabletop trashbin, pour balls into the bin.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9658822687933625,
        "cache_read_input_tokens": 2052608,
        "cache_reported_input_tokens": 2125112,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2125112,
        "cached_input_tokens": 2052608,
        "cost_usd": null,
        "fresh_input_tokens": 72504,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2125112,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2052608,
        "known_fresh_input_tokens": 72504,
        "known_input_tokens": 2125112,
        "known_output_tokens": 11030,
        "known_reasoning_output_tokens": 3397,
        "output_tokens": 11030,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3397,
        "reasoning_reported_output_tokens": 11030,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 63,
          "cache_write_input_tokens": 63,
          "cache_write_reported_input_tokens": 63,
          "cached_input_tokens": 63,
          "fresh_input_tokens": 63,
          "input_tokens": 63,
          "output_tokens": 63,
          "reasoning_output_tokens": 63,
          "reasoning_reported_output_tokens": 63
        },
        "request_attempts": 63,
        "response_count": 63,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2136142,
        "uncached_input_tokens": 72504,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11632,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11672,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11805,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12103,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12254,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 16764,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 18011,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18661,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 18891,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19082,
            "output_tokens": 173,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 22026,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22278,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22766,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 23124,
            "output_tokens": 217,
            "reasoning_output_tokens": 110
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 24061,
            "output_tokens": 189,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24485,
            "output_tokens": 337,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25320,
            "output_tokens": 612,
            "reasoning_output_tokens": 260
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 27186,
            "output_tokens": 126,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27778,
            "output_tokens": 1144,
            "reasoning_output_tokens": 328
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28943,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29106,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29932,
            "output_tokens": 92,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30051,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30877,
            "output_tokens": 59,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 32115,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 33977,
            "output_tokens": 124,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34128,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 35192,
            "output_tokens": 416,
            "reasoning_output_tokens": 279
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35845,
            "output_tokens": 130,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36278,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36470,
            "output_tokens": 78,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36852,
            "output_tokens": 124,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37282,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37475,
            "output_tokens": 183,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 37960,
            "output_tokens": 105,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38374,
            "output_tokens": 131,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38808,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39001,
            "output_tokens": 260,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39741,
            "output_tokens": 91,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 39859,
            "output_tokens": 170,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40648,
            "output_tokens": 180,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 40896,
            "output_tokens": 173,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41305,
            "output_tokens": 178,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41786,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42151,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42554,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42745,
            "output_tokens": 501,
            "reasoning_output_tokens": 156
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 44460,
            "output_tokens": 387,
            "reasoning_output_tokens": 296
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45067,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45056,
            "input_tokens": 45464,
            "output_tokens": 571,
            "reasoning_output_tokens": 468
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45952,
            "input_tokens": 46218,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46572,
            "output_tokens": 203,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 47079,
            "output_tokens": 52,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46976,
            "input_tokens": 47153,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 47344,
            "output_tokens": 39,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47552,
            "output_tokens": 366,
            "reasoning_output_tokens": 266
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 48140,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48128,
            "input_tokens": 48542,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 48904,
            "output_tokens": 715,
            "reasoning_output_tokens": 674
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49408,
            "input_tokens": 49681,
            "output_tokens": 366,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49920,
            "input_tokens": 50074,
            "output_tokens": 661,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50560,
            "input_tokens": 50757,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 70
      },
      "analysis": {
        "observed": "Native failure after 487 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-16/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-16/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-16/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-16/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-16/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-16/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-16/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-16/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-16/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-16/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-16/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-16/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-16/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-16/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-16/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-16/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-16/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-16/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/tabletop_motion.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-16/workspace/.kinex/tools/tabletop_motion.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/dump_bin.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-16/workspace/.kinex/memos/dump_bin.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 135,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 196,
        "published_frames": 196,
        "decode_ok": true,
        "duration_s": 4.9,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "54810423113f9fe1eb7fb1e1aa410393236eb5b1c0096146152854299ad144aa",
        "source_sha256": {
          "third_person": "4f956d13589ca331f3482cadaec3168b699d3ea884ab1f41e6c3c24c3c0c2190",
          "left_wrist": "e57d0b36c9a631103fd96b713c0c9f57453e07ab2fdb9ed888b1f2e291474f79",
          "right_wrist": "7cf4e8b5cbc743f5869a931df7efc3294d1db871d137cd5ded6b3a6b3c9f5039"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "8464f25e62ff",
          "task": "robotwin/dump-bin-bigbin-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 196,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 196,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19480000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 196,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 196,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19480000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 196,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 196,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19480000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-17",
      "variant": "kinex",
      "task_key": "task03/17",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "17",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 8.24,
        "step_limit": 400,
        "steps": 206,
        "success": true,
        "valid": true
      },
      "steps": 206,
      "simulation_time_s": 8.24,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 8.24,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use both arms to grab the light brown wooden roller.",
      "native_instruction": "Use both arms to grab the light brown wooden roller.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.958687654544358,
        "cache_read_input_tokens": 763008,
        "cache_reported_input_tokens": 795888,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 795888,
        "cached_input_tokens": 763008,
        "cost_usd": null,
        "fresh_input_tokens": 32880,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 795888,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 763008,
        "known_fresh_input_tokens": 32880,
        "known_input_tokens": 795888,
        "known_output_tokens": 4889,
        "known_reasoning_output_tokens": 626,
        "output_tokens": 4889,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 626,
        "reasoning_reported_output_tokens": 4889,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 36,
          "cache_write_input_tokens": 36,
          "cache_write_reported_input_tokens": 36,
          "cached_input_tokens": 36,
          "fresh_input_tokens": 36,
          "input_tokens": 36,
          "output_tokens": 36,
          "reasoning_output_tokens": 36,
          "reasoning_reported_output_tokens": 36
        },
        "request_attempts": 36,
        "response_count": 36,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 800777,
        "uncached_input_tokens": 32880,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11630,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11678,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11809,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12057,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12504,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 13318,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13184,
            "input_tokens": 14135,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14080,
            "input_tokens": 14961,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 15789,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15616,
            "input_tokens": 16581,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 16732,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 17980,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18210,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18400,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 18979,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 19231,
            "output_tokens": 173,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 20252,
            "output_tokens": 402,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 21283,
            "output_tokens": 121,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21870,
            "output_tokens": 163,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 23212,
            "output_tokens": 366,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24006,
            "output_tokens": 439,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24980,
            "output_tokens": 1025,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26027,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26902,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 28162,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28354,
            "output_tokens": 181,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 29108,
            "output_tokens": 62,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29232,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29425,
            "output_tokens": 151,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30155,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30674,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30905,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30999,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31696,
            "output_tokens": 252,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32072,
            "output_tokens": 487,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32580,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 35
      },
      "analysis": {
        "observed": "Native success after 206 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-17/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-17/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-17/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-17/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-17/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-17/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-17/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-17/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-17/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-17/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-17/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-17/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-17/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-17/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-17/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-17/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-17/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-17/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/dual_roller.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-17/workspace/.kinex/tools/dual_roller.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/roller.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-17/workspace/.kinex/memos/roller.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 73,
        "observed_images": 4
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 84,
        "published_frames": 84,
        "decode_ok": true,
        "duration_s": 2.1,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c0775338d6ad9e61011b8f7cabd5d0f1e04d4b39bc29fa6e0152a1ed09626f74",
        "source_sha256": {
          "third_person": "e6d938003e03cd4694c84714a4cc435d73a8f9a3ca7ee3d9726f2af13202a8cc",
          "left_wrist": "b1a0ab5b8b9e3e750cedb5b0a376fffe2d24d3cfb35cf00cadce6099955c3a7d",
          "right_wrist": "6ee91b1d60197688bb89a62928342e0cf824f388febf329b63ea3924bc98fdf2"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "485a6588b40f",
          "task": "robotwin/grab-roller-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 84,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 84,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8240000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 84,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 84,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8240000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 84,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 84,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8240000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-18",
      "variant": "kinex",
      "task_key": "task03/18",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "18",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 14.6,
        "step_limit": 600,
        "steps": 365,
        "success": false,
        "valid": true
      },
      "steps": 365,
      "simulation_time_s": 14.6,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 14.6,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the microphone used for audio input and pass it to the other arm.",
      "native_instruction": "Grab the microphone used for audio input and pass it to the other arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9689019977832941,
        "cache_read_input_tokens": 846208,
        "cache_reported_input_tokens": 873368,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 873368,
        "cached_input_tokens": 846208,
        "cost_usd": null,
        "fresh_input_tokens": 27160,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 873368,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 846208,
        "known_fresh_input_tokens": 27160,
        "known_input_tokens": 873368,
        "known_output_tokens": 5675,
        "known_reasoning_output_tokens": 1226,
        "output_tokens": 5675,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1226,
        "reasoning_reported_output_tokens": 5675,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 34,
          "cache_write_input_tokens": 34,
          "cache_write_reported_input_tokens": 34,
          "cached_input_tokens": 34,
          "fresh_input_tokens": 34,
          "input_tokens": 34,
          "output_tokens": 34,
          "reasoning_output_tokens": 34,
          "reasoning_reported_output_tokens": 34
        },
        "request_attempts": 34,
        "response_count": 34,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 879043,
        "uncached_input_tokens": 27160,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11634,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11674,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11807,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 13428,
            "output_tokens": 223,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13568,
            "input_tokens": 18083,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 19331,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 19561,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 19751,
            "output_tokens": 64,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 20355,
            "output_tokens": 125,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 21847,
            "output_tokens": 148,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22768,
            "output_tokens": 315,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 24035,
            "output_tokens": 316,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25100,
            "output_tokens": 1016,
            "reasoning_output_tokens": 271
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26137,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26254,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26974,
            "output_tokens": 69,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27070,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27696,
            "output_tokens": 51,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 27809,
            "output_tokens": 207,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28043,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28646,
            "output_tokens": 119,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29280,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29834,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30435,
            "output_tokens": 597,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31547,
            "output_tokens": 118,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 33205,
            "output_tokens": 175,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33878,
            "output_tokens": 248,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34336,
            "output_tokens": 196,
            "reasoning_output_tokens": 113
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 35030,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35582,
            "output_tokens": 72,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36153,
            "output_tokens": 116,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36767,
            "output_tokens": 671,
            "reasoning_output_tokens": 82
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37463,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 42
      },
      "analysis": {
        "observed": "Native failure after 365 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-18/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-18/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-18/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-18/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-18/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-18/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-18/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-18/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-18/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-18/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-18/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-18/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-18/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-18/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-18/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-18/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-18/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-18/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-18/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/handover-microphone.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-18/workspace/.kinex/memos/handover-microphone.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 78,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 147,
        "published_frames": 147,
        "decode_ok": true,
        "duration_s": 3.675,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c5878dba5e1e9bb9d6fc9023c4b513efe748a8c526a15f2b844962c568da9228",
        "source_sha256": {
          "third_person": "37124a9f6cd5bc2fa7d2209e2a1ecfabef29d7d5465ff06c4dadf31279562795",
          "left_wrist": "95695f880a97be66529bf688853b0e15015271afc72066738ec1fb5eb76af7fd",
          "right_wrist": "570b9c9425c97a67d33f90b67b8a82a8556b8bb604c91a94350e1ca8a2d2e98e"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "c208cecbd942",
          "task": "robotwin/handover-mic-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 148,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 147,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 148,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 147,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 148,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 147,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-19",
      "variant": "kinex",
      "task_key": "task03/19",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "19",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 10.0,
        "step_limit": 400,
        "steps": 250,
        "success": true,
        "valid": true
      },
      "steps": 250,
      "simulation_time_s": 10.0,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 10.0,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the medium sauce can and place it near the smooth gray pot with matte finish",
      "native_instruction": "Grab the medium sauce can and place it near the smooth gray pot with matte finish",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9294790445368545,
        "cache_read_input_tokens": 635008,
        "cache_reported_input_tokens": 683187,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 683187,
        "cached_input_tokens": 635008,
        "cost_usd": null,
        "fresh_input_tokens": 48179,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 683187,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 635008,
        "known_fresh_input_tokens": 48179,
        "known_input_tokens": 683187,
        "known_output_tokens": 5184,
        "known_reasoning_output_tokens": 1011,
        "output_tokens": 5184,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1011,
        "reasoning_reported_output_tokens": 5184,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 30,
          "cache_write_input_tokens": 30,
          "cache_write_reported_input_tokens": 30,
          "cached_input_tokens": 30,
          "fresh_input_tokens": 30,
          "input_tokens": 30,
          "output_tokens": 30,
          "reasoning_output_tokens": 30,
          "reasoning_reported_output_tokens": 30
        },
        "request_attempts": 30,
        "response_count": 30,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 688371,
        "uncached_input_tokens": 48179,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11637,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11677,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11810,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11860,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 12108,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12555,
            "output_tokens": 165,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 15947,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 17194,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 17424,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 17615,
            "output_tokens": 195,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 20577,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20829,
            "output_tokens": 310,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 22531,
            "output_tokens": 511,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23346,
            "output_tokens": 93,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23905,
            "output_tokens": 555,
            "reasoning_output_tokens": 218
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24887,
            "output_tokens": 392,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25759,
            "output_tokens": 959,
            "reasoning_output_tokens": 120
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26739,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27464,
            "output_tokens": 140,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27631,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28096,
            "output_tokens": 96,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28561,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 28981,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29420,
            "output_tokens": 195,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29985,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30421,
            "output_tokens": 73,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30626,
            "output_tokens": 97,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30785,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31087,
            "output_tokens": 617,
            "reasoning_output_tokens": 86
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31730,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 36
      },
      "analysis": {
        "observed": "Native success after 250 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-19/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-19/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-19/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-19/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-19/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-19/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-19/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-19/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-19/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-19/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-19/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-19/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-19/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-19/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-19/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-19/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-19/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-19/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/tcp_motion.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-19/workspace/.kinex/tools/tcp_motion.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/aloha-table-manipulation.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-19/workspace/.kinex/memos/aloha-table-manipulation.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 68,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 101,
        "published_frames": 101,
        "decode_ok": true,
        "duration_s": 2.525,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "4779d66a35464bb1b488f5465e1a50eda073a68f5609470264ad7cf893f87a77",
        "source_sha256": {
          "third_person": "4422cfb4d8c31ee560f53d5dadb8c85906e3e3362c58b870e6c3a31631d78a64",
          "left_wrist": "97a4ab25d9ba4b1d27f09b0a79459032e04797949e0ba2aa133db3df3bd0abf7",
          "right_wrist": "c64653488d87a5190cfef96f14247c780275c296f864c8e1db31ff981c58607e"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ebbb7703c781",
          "task": "robotwin/move-can-pot-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 102,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 101,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10000000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 102,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 101,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10000000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 102,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 101,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10000000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-20",
      "variant": "kinex",
      "task_key": "task03/20",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "20",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 11.92,
        "step_limit": 400,
        "steps": 298,
        "success": true,
        "valid": true
      },
      "steps": 298,
      "simulation_time_s": 11.92,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 11.92,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the medium-sized bottle and put it onto the pad",
      "native_instruction": "Pick the medium-sized bottle and put it onto the pad",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9748120363667184,
        "cache_read_input_tokens": 1805696,
        "cache_reported_input_tokens": 1852353,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1852353,
        "cached_input_tokens": 1805696,
        "cost_usd": null,
        "fresh_input_tokens": 46657,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1852353,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1805696,
        "known_fresh_input_tokens": 46657,
        "known_input_tokens": 1852353,
        "known_output_tokens": 8647,
        "known_reasoning_output_tokens": 2761,
        "output_tokens": 8647,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2761,
        "reasoning_reported_output_tokens": 8647,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 53,
          "cache_write_input_tokens": 53,
          "cache_write_reported_input_tokens": 53,
          "cached_input_tokens": 53,
          "fresh_input_tokens": 53,
          "input_tokens": 53,
          "output_tokens": 53,
          "reasoning_output_tokens": 53,
          "reasoning_reported_output_tokens": 53
        },
        "request_attempts": 53,
        "response_count": 53,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1861000,
        "uncached_input_tokens": 46657,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11636,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11676,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11809,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 15336,
            "output_tokens": 246,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15488,
            "input_tokens": 19258,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 20506,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20736,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 20927,
            "output_tokens": 182,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 23952,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24440,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24799,
            "output_tokens": 189,
            "reasoning_output_tokens": 60
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25892,
            "output_tokens": 279,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 27123,
            "output_tokens": 282,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 28372,
            "output_tokens": 246,
            "reasoning_output_tokens": 204
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 29797,
            "output_tokens": 666,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30484,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30702,
            "output_tokens": 83,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31480,
            "output_tokens": 181,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 32344,
            "output_tokens": 166,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 33194,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 35056,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35157,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35959,
            "output_tokens": 162,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36257,
            "output_tokens": 142,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36634,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 36827,
            "output_tokens": 224,
            "reasoning_output_tokens": 127
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37268,
            "output_tokens": 565,
            "reasoning_output_tokens": 448
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 37968,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38016,
            "input_tokens": 38303,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38661,
            "output_tokens": 264,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39333,
            "output_tokens": 377,
            "reasoning_output_tokens": 285
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40462,
            "output_tokens": 185,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40576,
            "input_tokens": 40674,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41489,
            "output_tokens": 190,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41706,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42027,
            "output_tokens": 67,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42263,
            "output_tokens": 423,
            "reasoning_output_tokens": 284
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42929,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43293,
            "output_tokens": 71,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43386,
            "output_tokens": 153,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43644,
            "output_tokens": 70,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 43883,
            "output_tokens": 140,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44142,
            "output_tokens": 117,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44499,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44693,
            "output_tokens": 474,
            "reasoning_output_tokens": 242
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45269,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45604,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45568,
            "input_tokens": 45940,
            "output_tokens": 82,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46193,
            "output_tokens": 152,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 46479,
            "output_tokens": 68,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 46776,
            "output_tokens": 496,
            "reasoning_output_tokens": 176
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47299,
            "output_tokens": 493,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 47817,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 65
      },
      "analysis": {
        "observed": "Native success after 298 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-20/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-20/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-20/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-20/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-20/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-20/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-20/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-20/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-20/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-20/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-20/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-20/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-20/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-20/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-20/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-20/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-20/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-20/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_step.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-20/workspace/.kinex/tools/cartesian_step.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-20/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 120,
        "observed_images": 13
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 120,
        "published_frames": 120,
        "decode_ok": true,
        "duration_s": 3.0,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "d60b28b04246d6a7eada53238b9648e58a51de6e239301f61c79e2fec3751af3",
        "source_sha256": {
          "third_person": "ced73b1076008cbc1ff27bfe9ffb8bf3011faecec3a60d52cfb0109a9d4bf918",
          "left_wrist": "7ef5257210d5041365e46125927985bc90e8d4a3fd08d74ed035df5bc7acd4fe",
          "right_wrist": "dc1e779ae3ccf67f91632e45c34877f5225623ae2951b454650e27c6723e634f"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "79e6201bf0f8",
          "task": "robotwin/move-pillbottle-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 121,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 120,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11920000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 121,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 120,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11920000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 121,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 120,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11920000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-21",
      "variant": "kinex",
      "task_key": "task03/21",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "21",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 10.52,
        "step_limit": 400,
        "steps": 263,
        "success": true,
        "valid": true
      },
      "steps": 263,
      "simulation_time_s": 10.52,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 10.52,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the small packet of playingcards and move it outward.",
      "native_instruction": "Pick up the small packet of playingcards and move it outward.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9477986182174115,
        "cache_read_input_tokens": 596480,
        "cache_reported_input_tokens": 629332,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 629332,
        "cached_input_tokens": 596480,
        "cost_usd": null,
        "fresh_input_tokens": 32852,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 629332,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 596480,
        "known_fresh_input_tokens": 32852,
        "known_input_tokens": 629332,
        "known_output_tokens": 4554,
        "known_reasoning_output_tokens": 842,
        "output_tokens": 4554,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 842,
        "reasoning_reported_output_tokens": 4554,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 28,
          "cache_write_input_tokens": 28,
          "cache_write_reported_input_tokens": 28,
          "cached_input_tokens": 28,
          "fresh_input_tokens": 28,
          "input_tokens": 28,
          "output_tokens": 28,
          "reasoning_output_tokens": 28,
          "reasoning_reported_output_tokens": 28
        },
        "request_attempts": 28,
        "response_count": 28,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 633886,
        "uncached_input_tokens": 32852,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11633,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11673,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11806,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 12105,
            "output_tokens": 223,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 16760,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 18007,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18237,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 18427,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 19006,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 19258,
            "output_tokens": 418,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 21228,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21499,
            "output_tokens": 290,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 23092,
            "output_tokens": 1081,
            "reasoning_output_tokens": 179
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24194,
            "output_tokens": 238,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24605,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25314,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25433,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25874,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26329,
            "output_tokens": 80,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26431,
            "output_tokens": 80,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26876,
            "output_tokens": 127,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27371,
            "output_tokens": 342,
            "reasoning_output_tokens": 253
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28074,
            "output_tokens": 308,
            "reasoning_output_tokens": 187
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28743,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28928,
            "output_tokens": 67,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29017,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29463,
            "output_tokens": 461,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 29949,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 35
      },
      "analysis": {
        "observed": "Native success after 263 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-21/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-21/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-21/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-21/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-21/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-21/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-21/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-21/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-21/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-21/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-21/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-21/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-21/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-21/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-21/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-21/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-21/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-21/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-21/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/aloha_manipulation.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-21/workspace/.kinex/memos/aloha_manipulation.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 65,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 106,
        "published_frames": 106,
        "decode_ok": true,
        "duration_s": 2.65,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c7e445975ab22e353e35efa0137bdcd95ea908a8cac5f3c51f8bb0b3e73477ea",
        "source_sha256": {
          "third_person": "096f56f052d6ca050c8afec4d6efe0a7d5f5cafd08da9946e9d765df9970ab16",
          "left_wrist": "ca046f3ccc1a00a0e8c56ecde5215601ffda08bacb1a03992c10a47125333380",
          "right_wrist": "21a7c809e0aa0b5fc924b0e2e6f97e49dc878276ba4e9588fb5aaafa88aa1331"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "cbc796d55563",
          "task": "robotwin/move-playingcard-away-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 107,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 106,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10520000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 107,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 106,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10520000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 107,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 106,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10520000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-22",
      "variant": "kinex",
      "task_key": "task03/22",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "22",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 11.200000000000001,
        "step_limit": 400,
        "steps": 280,
        "success": false,
        "valid": true
      },
      "steps": 280,
      "simulation_time_s": 11.200000000000001,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 11.200000000000001,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the red plastic stapler and set it on Red mat",
      "native_instruction": "Grab the red plastic stapler and set it on Red mat",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9697536406670881,
        "cache_read_input_tokens": 912640,
        "cache_reported_input_tokens": 941105,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 941105,
        "cached_input_tokens": 912640,
        "cost_usd": null,
        "fresh_input_tokens": 28465,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 941105,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 912640,
        "known_fresh_input_tokens": 28465,
        "known_input_tokens": 941105,
        "known_output_tokens": 5224,
        "known_reasoning_output_tokens": 722,
        "output_tokens": 5224,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 722,
        "reasoning_reported_output_tokens": 5224,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 35,
          "cache_write_input_tokens": 35,
          "cache_write_reported_input_tokens": 35,
          "cached_input_tokens": 35,
          "fresh_input_tokens": 35,
          "input_tokens": 35,
          "output_tokens": 35,
          "reasoning_output_tokens": 35,
          "reasoning_reported_output_tokens": 35
        },
        "request_attempts": 35,
        "response_count": 35,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 946329,
        "uncached_input_tokens": 28465,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11632,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11672,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11805,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11855,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 13324,
            "output_tokens": 205,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13440,
            "input_tokens": 17169,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17024,
            "input_tokens": 19684,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 19914,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 20104,
            "output_tokens": 222,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 23629,
            "output_tokens": 81,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 23920,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24211,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 25049,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25320,
            "output_tokens": 283,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 26138,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26562,
            "output_tokens": 294,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27461,
            "output_tokens": 126,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28053,
            "output_tokens": 364,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 29436,
            "output_tokens": 137,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30752,
            "output_tokens": 695,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31470,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31629,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31926,
            "output_tokens": 157,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32446,
            "output_tokens": 172,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32990,
            "output_tokens": 94,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33437,
            "output_tokens": 92,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33551,
            "output_tokens": 88,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33994,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34460,
            "output_tokens": 197,
            "reasoning_output_tokens": 96
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35025,
            "output_tokens": 150,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35542,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35945,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36416,
            "output_tokens": 398,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 36996,
            "output_tokens": 566,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37588,
            "output_tokens": 18,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 43
      },
      "analysis": {
        "observed": "Native failure after 280 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-22/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-22/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-22/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-22/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-22/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-22/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-22/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-22/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-22/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-22/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-22/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-22/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-22/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-22/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-22/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-22/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-22/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-22/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_step.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-22/workspace/.kinex/tools/cartesian_step.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/aloha-tabletop.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-22/workspace/.kinex/memos/aloha-tabletop.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 80,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 113,
        "published_frames": 113,
        "decode_ok": true,
        "duration_s": 2.825,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c24b918c1a4e74f8c209984749912c8d12f4f537ce2b0873db1f55c222a80024",
        "source_sha256": {
          "third_person": "3b86097800dc6956f30026f99b3bfb9647368120c0d3beaf597b0d7a949a5176",
          "left_wrist": "0f1d4652daa3f532290f6106dd1e1636a038d20463757e2508304b8b6777c49d",
          "right_wrist": "38e7edede20a262d9b8fa48950743eb63bcf2e69da2f3918725f8ae36b281591"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "e4db09c90d35",
          "task": "robotwin/move-stapler-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 114,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 113,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 114,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 113,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 114,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 113,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 11200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-23",
      "variant": "kinex",
      "task_key": "task03/23",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "23",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.040000000000001,
        "step_limit": 700,
        "steps": 226,
        "success": true,
        "valid": true
      },
      "steps": 226,
      "simulation_time_s": 9.040000000000001,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 9.040000000000001,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift the black laptop using the right arm.",
      "native_instruction": "Lift the black laptop using the right arm.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9604015242192424,
        "cache_read_input_tokens": 830464,
        "cache_reported_input_tokens": 864705,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 864705,
        "cached_input_tokens": 830464,
        "cost_usd": null,
        "fresh_input_tokens": 34241,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 864705,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 830464,
        "known_fresh_input_tokens": 34241,
        "known_input_tokens": 864705,
        "known_output_tokens": 6348,
        "known_reasoning_output_tokens": 957,
        "output_tokens": 6348,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 957,
        "reasoning_reported_output_tokens": 6348,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 34,
          "cache_write_input_tokens": 34,
          "cache_write_reported_input_tokens": 34,
          "cached_input_tokens": 34,
          "fresh_input_tokens": 34,
          "input_tokens": 34,
          "output_tokens": 34,
          "reasoning_output_tokens": 34,
          "reasoning_reported_output_tokens": 34
        },
        "request_attempts": 34,
        "response_count": 34,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 871053,
        "uncached_input_tokens": 34241,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11631,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11671,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11804,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 13789,
            "output_tokens": 215,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13824,
            "input_tokens": 16905,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16768,
            "input_tokens": 18153,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18317,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18176,
            "input_tokens": 18775,
            "output_tokens": 107,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 19422,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 19674,
            "output_tokens": 301,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 20370,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 20641,
            "output_tokens": 290,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 22059,
            "output_tokens": 334,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22843,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23718,
            "output_tokens": 296,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24367,
            "output_tokens": 206,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 26501,
            "output_tokens": 815,
            "reasoning_output_tokens": 145
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27339,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27647,
            "output_tokens": 129,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28955,
            "output_tokens": 975,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 29953,
            "output_tokens": 125,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30174,
            "output_tokens": 141,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 31000,
            "output_tokens": 123,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31150,
            "output_tokens": 141,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31433,
            "output_tokens": 141,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31994,
            "output_tokens": 195,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32616,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33083,
            "output_tokens": 376,
            "reasoning_output_tokens": 175
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33808,
            "output_tokens": 155,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34382,
            "output_tokens": 196,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34716,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34784,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35230,
            "output_tokens": 549,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35801,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 42
      },
      "analysis": {
        "observed": "Native success after 226 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-23/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-23/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-23/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-23/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-23/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-23/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-23/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-23/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-23/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-23/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-23/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-23/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-23/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-23/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-23/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-23/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-23/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-23/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/laptop_geometry.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-23/workspace/.kinex/tools/laptop_geometry.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/right_tcp_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-23/workspace/.kinex/tools/right_tcp_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/open-laptop.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-23/workspace/.kinex/memos/open-laptop.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 78,
        "observed_images": 7
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 92,
        "published_frames": 92,
        "decode_ok": true,
        "duration_s": 2.3,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "e27fdc11e900cab5141476890d4975df90ca4eafdd74b2f9b30189e5cabfc4c7",
        "source_sha256": {
          "third_person": "4593f0417fcab27c4d7fd58dc6347cedb0be2a70ca19624f9d73319c0d2829ca",
          "left_wrist": "2c6e7e841cd6c9d64a87ef81bf0a9ea2b2170b5457fc32d4ef97a1c21a5e4cd3",
          "right_wrist": "1d37c345e5ca15fd007f44f65c171834e396a2153520a4beadfcff70d0e6b324"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "8667761c2705",
          "task": "robotwin/open-laptop-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 92,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 92,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9040000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 92,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 92,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9040000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 92,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 92,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9040000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-24",
      "variant": "kinex",
      "task_key": "task03/24",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "24",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.24,
        "step_limit": 1500,
        "steps": 306,
        "success": true,
        "valid": true
      },
      "steps": 306,
      "simulation_time_s": 12.24,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 12.24,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to pull the handle of the kitchen microwave with slanted shape.",
      "native_instruction": "Use the left arm to pull the handle of the kitchen microwave with slanted shape.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9536469504049221,
        "cache_read_input_tokens": 575360,
        "cache_reported_input_tokens": 603326,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 603326,
        "cached_input_tokens": 575360,
        "cost_usd": null,
        "fresh_input_tokens": 27966,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 603326,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 575360,
        "known_fresh_input_tokens": 27966,
        "known_input_tokens": 603326,
        "known_output_tokens": 4246,
        "known_reasoning_output_tokens": 861,
        "output_tokens": 4246,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 861,
        "reasoning_reported_output_tokens": 4246,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 31,
          "cache_write_input_tokens": 31,
          "cache_write_reported_input_tokens": 31,
          "cached_input_tokens": 31,
          "fresh_input_tokens": 31,
          "input_tokens": 31,
          "output_tokens": 31,
          "reasoning_output_tokens": 31,
          "reasoning_reported_output_tokens": 31
        },
        "request_attempts": 31,
        "response_count": 31,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 607572,
        "uncached_input_tokens": 27966,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11637,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11768,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11816,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12064,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12511,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 13325,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13184,
            "input_tokens": 14151,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14080,
            "input_tokens": 14943,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 15094,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14976,
            "input_tokens": 16341,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16256,
            "input_tokens": 16505,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 16960,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 17539,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 17791,
            "output_tokens": 63,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 18320,
            "output_tokens": 295,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 19010,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 19281,
            "output_tokens": 285,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 20857,
            "output_tokens": 314,
            "reasoning_output_tokens": 239
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21866,
            "output_tokens": 1006,
            "reasoning_output_tokens": 107
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 22893,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23344,
            "output_tokens": 175,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23894,
            "output_tokens": 61,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 23977,
            "output_tokens": 46,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24794,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24835,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25028,
            "output_tokens": 119,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25513,
            "output_tokens": 210,
            "reasoning_output_tokens": 117
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26101,
            "output_tokens": 307,
            "reasoning_output_tokens": 135
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26542,
            "output_tokens": 74,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27011,
            "output_tokens": 581,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27615,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 30
      },
      "analysis": {
        "observed": "Native success after 306 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-24/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-24/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-24/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-24/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-24/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-24/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-24/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-24/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-24/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-24/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-24/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-24/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-24/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-24/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-24/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-24/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-24/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-24/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_left.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-24/workspace/.kinex/tools/cartesian_left.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/open-microwave.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-24/workspace/.kinex/memos/open-microwave.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 63,
        "observed_images": 7
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1500,
        "max_control_steps": 1500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 124,
        "published_frames": 124,
        "decode_ok": true,
        "duration_s": 3.1,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c6bb3147e8c9a80a8aecd3966f38e1caedfe861f493153a1bc623014f6b851e7",
        "source_sha256": {
          "third_person": "e78bc2bef40b0dc0dd9ec58db73974515f653ec3a2449ced5447e676ea2badfa",
          "left_wrist": "6f3e7b633a41415dd68830dd434ecc3b6a2858f85d9db86a360fd6acd83d9c22",
          "right_wrist": "f7bc5efd2d511420fbeb1a1b5e614236e3073376d0014f4faef5fe67315dacc0"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "6a1e52a18516",
          "task": "robotwin/open-microwave-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 124,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 124,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12240000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 124,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 124,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12240000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 124,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 124,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12240000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-25",
      "variant": "kinex",
      "task_key": "task03/25",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "25",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 2
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.96,
        "step_limit": 400,
        "steps": 399,
        "success": false,
        "valid": true
      },
      "steps": 399,
      "simulation_time_s": 15.96,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 15.96,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift the bottle with attached loop on cap and the plastic bottle with blue label simultaneously with separate arms.",
      "native_instruction": "Lift the bottle with attached loop on cap and the plastic bottle with blue label simultaneously with separate arms.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9759429456745765,
        "cache_read_input_tokens": 1800320,
        "cache_reported_input_tokens": 1844698,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1844698,
        "cached_input_tokens": 1800320,
        "cost_usd": null,
        "fresh_input_tokens": 44378,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1844698,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1800320,
        "known_fresh_input_tokens": 44378,
        "known_input_tokens": 1844698,
        "known_output_tokens": 10764,
        "known_reasoning_output_tokens": 3924,
        "output_tokens": 10764,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3924,
        "reasoning_reported_output_tokens": 10764,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 56,
          "cache_write_input_tokens": 56,
          "cache_write_reported_input_tokens": 56,
          "cached_input_tokens": 56,
          "fresh_input_tokens": 56,
          "input_tokens": 56,
          "output_tokens": 56,
          "reasoning_output_tokens": 56,
          "reasoning_reported_output_tokens": 56
        },
        "request_attempts": 56,
        "response_count": 56,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1855462,
        "uncached_input_tokens": 44378,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11647,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11687,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11820,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11870,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 13607,
            "output_tokens": 205,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13696,
            "input_tokens": 17452,
            "output_tokens": 178,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 19767,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 21028,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 21258,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21449,
            "output_tokens": 305,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 23343,
            "output_tokens": 152,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 24224,
            "output_tokens": 273,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24957,
            "output_tokens": 595,
            "reasoning_output_tokens": 180
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25886,
            "output_tokens": 185,
            "reasoning_output_tokens": 143
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 27250,
            "output_tokens": 1072,
            "reasoning_output_tokens": 118
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28346,
            "output_tokens": 110,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28662,
            "output_tokens": 71,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29428,
            "output_tokens": 124,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30103,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30296,
            "output_tokens": 163,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31016,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31210,
            "output_tokens": 144,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31416,
            "output_tokens": 418,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32234,
            "output_tokens": 254,
            "reasoning_output_tokens": 148
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32962,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33156,
            "output_tokens": 519,
            "reasoning_output_tokens": 165
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34157,
            "output_tokens": 223,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34706,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35121,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35315,
            "output_tokens": 111,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35821,
            "output_tokens": 116,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36106,
            "output_tokens": 78,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36206,
            "output_tokens": 45,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36269,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36461,
            "output_tokens": 319,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37106,
            "output_tokens": 568,
            "reasoning_output_tokens": 470
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38369,
            "output_tokens": 1023,
            "reasoning_output_tokens": 924
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39717,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40129,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40322,
            "output_tokens": 38,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40192,
            "input_tokens": 40530,
            "output_tokens": 85,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 41012,
            "output_tokens": 543,
            "reasoning_output_tokens": 176
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41720,
            "output_tokens": 272,
            "reasoning_output_tokens": 172
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42319,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42776,
            "output_tokens": 96,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 43042,
            "output_tokens": 47,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43259,
            "output_tokens": 211,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 44028,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44222,
            "output_tokens": 158,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44431,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44416,
            "input_tokens": 44623,
            "output_tokens": 292,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45140,
            "output_tokens": 338,
            "reasoning_output_tokens": 213
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 46034,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45952,
            "input_tokens": 46226,
            "output_tokens": 183,
            "reasoning_output_tokens": 150
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46471,
            "output_tokens": 488,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46848,
            "input_tokens": 46986,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 67
      },
      "analysis": {
        "observed": "Native failure after 399 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-25/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-25/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-25/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-25/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-25/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-25/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-25/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-25/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-25/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-25/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-25/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-25/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-25/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-25/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-25/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-25/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-25/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-25/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/dual_cartesian.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-25/workspace/.kinex/tools/dual_cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin_bottles.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-25/workspace/.kinex/memos/robotwin_bottles.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 125,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 161,
        "published_frames": 161,
        "decode_ok": true,
        "duration_s": 4.025,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "6b0c56f0d07ad34611f79fd34406ccc0c9037531fbe71e7806d7fbd5057da04e",
        "source_sha256": {
          "third_person": "8a2c6591029e8931585aec67316dd74aeb55639be4d0df73a607231f168796d6",
          "left_wrist": "7b886220d70ab2c4863023e10c91da11beb36128dfedd472111abbb5bab0b3bc",
          "right_wrist": "62d4225037464c0bef1ca9faf48b419134c8fd9d3099782057ac55965236443c"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "7253c1d12c1e",
          "task": "robotwin/pick-diverse-bottles-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-26",
      "variant": "kinex",
      "task_key": "task03/26",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "26",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 9.200000000000001,
        "step_limit": 400,
        "steps": 230,
        "success": false,
        "valid": true
      },
      "steps": 230,
      "simulation_time_s": 9.200000000000001,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 9.200000000000001,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use one arm for the bottle with red screw cap and the other for the green bottle.",
      "native_instruction": "Use one arm for the bottle with red screw cap and the other for the green bottle.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9624702901255181,
        "cache_read_input_tokens": 734976,
        "cache_reported_input_tokens": 763635,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 763635,
        "cached_input_tokens": 734976,
        "cost_usd": null,
        "fresh_input_tokens": 28659,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 763635,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 734976,
        "known_fresh_input_tokens": 28659,
        "known_input_tokens": 763635,
        "known_output_tokens": 5653,
        "known_reasoning_output_tokens": 784,
        "output_tokens": 5653,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 784,
        "reasoning_reported_output_tokens": 5653,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 28,
          "cache_write_input_tokens": 28,
          "cache_write_reported_input_tokens": 28,
          "cached_input_tokens": 28,
          "fresh_input_tokens": 28,
          "input_tokens": 28,
          "output_tokens": 28,
          "reasoning_output_tokens": 28,
          "reasoning_reported_output_tokens": 28
        },
        "request_attempts": 28,
        "response_count": 28,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 769288,
        "uncached_input_tokens": 28659,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11643,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11678,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11809,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 15618,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15488,
            "input_tokens": 15683,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 19516,
            "output_tokens": 168,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 22900,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23130,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23321,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23573,
            "output_tokens": 426,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24394,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24665,
            "output_tokens": 377,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 27219,
            "output_tokens": 673,
            "reasoning_output_tokens": 212
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28216,
            "output_tokens": 128,
            "reasoning_output_tokens": 86
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 29523,
            "output_tokens": 1225,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30770,
            "output_tokens": 136,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30934,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31687,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31764,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32193,
            "output_tokens": 200,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 33462,
            "output_tokens": 229,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 34657,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35049,
            "output_tokens": 233,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 36253,
            "output_tokens": 152,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 37945,
            "output_tokens": 77,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38201,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38563,
            "output_tokens": 682,
            "reasoning_output_tokens": 110
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39269,
            "output_tokens": 43,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 36
      },
      "analysis": {
        "observed": "Native failure after 230 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-26/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-26/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-26/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-26/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-26/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-26/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-26/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-26/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-26/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-26/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-26/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-26/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-26/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-26/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-26/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-26/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-26/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-26/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/dual_cartesian.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-26/workspace/.kinex/tools/dual_cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/dual_bottle_pick.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-26/workspace/.kinex/memos/dual_bottle_pick.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 66,
        "observed_images": 7
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 93,
        "published_frames": 93,
        "decode_ok": true,
        "duration_s": 2.325,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c53269ca072d26c96754480211c1df3345a8a446383374e12c94fa75df64eee3",
        "source_sha256": {
          "third_person": "db1528feaf425a7dac152a424b4dca0da353262d5b81e83395a0fe166e69bab3",
          "left_wrist": "9d04e1bc57789bd9a8cf158a1692460cd00c1fefefdc8aaa401f9b5466ccf749",
          "right_wrist": "121c41d605257648c5b4894de04571fe9b7ee45a65be098520583cb4f979fdca"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "b881ced49648",
          "task": "robotwin/pick-dual-bottles-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 94,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 93,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 94,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 93,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 94,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 93,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-27",
      "variant": "kinex",
      "task_key": "task03/27",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "27",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 7.96,
        "step_limit": 400,
        "steps": 199,
        "success": true,
        "valid": true
      },
      "steps": 199,
      "simulation_time_s": 7.96,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 7.96,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Position the coffee-box with mug image on the left of the mouse with scroll wheel",
      "native_instruction": "Position the coffee-box with mug image on the left of the mouse with scroll wheel",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9504544872615542,
        "cache_read_input_tokens": 527104,
        "cache_reported_input_tokens": 554581,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 554581,
        "cached_input_tokens": 527104,
        "cost_usd": null,
        "fresh_input_tokens": 27477,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 554581,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 527104,
        "known_fresh_input_tokens": 27477,
        "known_input_tokens": 554581,
        "known_output_tokens": 4139,
        "known_reasoning_output_tokens": 592,
        "output_tokens": 4139,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 592,
        "reasoning_reported_output_tokens": 4139,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 27,
          "cache_write_input_tokens": 27,
          "cache_write_reported_input_tokens": 27,
          "cached_input_tokens": 27,
          "fresh_input_tokens": 27,
          "input_tokens": 27,
          "output_tokens": 27,
          "reasoning_output_tokens": 27,
          "reasoning_reported_output_tokens": 27
        },
        "request_attempts": 27,
        "response_count": 27,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 558720,
        "uncached_input_tokens": 27477,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11640,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11675,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11808,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11856,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12104,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 15126,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14976,
            "input_tokens": 16374,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16256,
            "input_tokens": 16538,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 16994,
            "output_tokens": 85,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 17619,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 17871,
            "output_tokens": 135,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18929,
            "output_tokens": 251,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 19576,
            "output_tokens": 179,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 22091,
            "output_tokens": 503,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22894,
            "output_tokens": 355,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23397,
            "output_tokens": 939,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24357,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24761,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 25170,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 25438,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25853,
            "output_tokens": 202,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26403,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26812,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26991,
            "output_tokens": 46,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27075,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27342,
            "output_tokens": 520,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 27887,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 32
      },
      "analysis": {
        "observed": "Native success after 199 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-27/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-27/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-27/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-27/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-27/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-27/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-27/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-27/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-27/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-27/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-27/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-27/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-27/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-27/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-27/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-27/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-27/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-27/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-27/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/aloha_manipulation.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-27/workspace/.kinex/memos/aloha_manipulation.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 61,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 81,
        "published_frames": 81,
        "decode_ok": true,
        "duration_s": 2.025,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "ca851b879f622c3fdb458a095dc691ca5a4082a4343559c7a6f197e6c2a78b28",
        "source_sha256": {
          "third_person": "407ba627db2db458a7ed662be65ef73259fa71dd4e41d5a0724bdedbb8752459",
          "left_wrist": "f2d329223438a9c9a4a4e27176296aa61c56cdfc5d622360275a5c9729fe9dbb",
          "right_wrist": "32198171005f674bc3c1b57fb205d8911bbe0af34a120fbb250702eba0324bb8"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "1a93af0f9bea",
          "task": "robotwin/place-a2b-left-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 81,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 81,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 81,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 81,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 81,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 81,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 7960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-28",
      "variant": "kinex",
      "task_key": "task03/28",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "28",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 15.96,
        "step_limit": 400,
        "steps": 399,
        "success": false,
        "valid": true
      },
      "steps": 399,
      "simulation_time_s": 15.96,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 15.96,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Put the tiny pink car with four wheels to the right of the red, blue, green, yellow rubikscube",
      "native_instruction": "Put the tiny pink car with four wheels to the right of the red, blue, green, yellow rubikscube",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9652789714910078,
        "cache_read_input_tokens": 1134336,
        "cache_reported_input_tokens": 1175138,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1175138,
        "cached_input_tokens": 1134336,
        "cost_usd": null,
        "fresh_input_tokens": 40802,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1175138,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1134336,
        "known_fresh_input_tokens": 40802,
        "known_input_tokens": 1175138,
        "known_output_tokens": 7136,
        "known_reasoning_output_tokens": 2631,
        "output_tokens": 7136,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2631,
        "reasoning_reported_output_tokens": 7136,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 41,
          "cache_write_input_tokens": 41,
          "cache_write_reported_input_tokens": 41,
          "cached_input_tokens": 41,
          "fresh_input_tokens": 41,
          "input_tokens": 41,
          "output_tokens": 41,
          "reasoning_output_tokens": 41,
          "reasoning_reported_output_tokens": 41
        },
        "request_attempts": 41,
        "response_count": 41,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1182274,
        "uncached_input_tokens": 40802,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11646,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11686,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11819,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 12118,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12267,
            "output_tokens": 162,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 15294,
            "output_tokens": 175,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15360,
            "input_tokens": 19114,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 19408,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 20326,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 20912,
            "output_tokens": 111,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21233,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21996,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22267,
            "output_tokens": 243,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 23568,
            "output_tokens": 52,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23943,
            "output_tokens": 160,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 26439,
            "output_tokens": 302,
            "reasoning_output_tokens": 199
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26943,
            "output_tokens": 759,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 27723,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28485,
            "output_tokens": 101,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28792,
            "output_tokens": 123,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29435,
            "output_tokens": 238,
            "reasoning_output_tokens": 152
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30196,
            "output_tokens": 232,
            "reasoning_output_tokens": 86
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30623,
            "output_tokens": 305,
            "reasoning_output_tokens": 177
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31457,
            "output_tokens": 131,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31739,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32329,
            "output_tokens": 488,
            "reasoning_output_tokens": 408
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33343,
            "output_tokens": 94,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33964,
            "output_tokens": 276,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 35489,
            "output_tokens": 324,
            "reasoning_output_tokens": 247
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36336,
            "output_tokens": 255,
            "reasoning_output_tokens": 173
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37122,
            "output_tokens": 390,
            "reasoning_output_tokens": 308
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 38204,
            "output_tokens": 96,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38823,
            "output_tokens": 283,
            "reasoning_output_tokens": 207
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39630,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39822,
            "output_tokens": 190,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40537,
            "output_tokens": 315,
            "reasoning_output_tokens": 213
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41377,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41565,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 42166,
            "output_tokens": 47,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42275,
            "output_tokens": 427,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 42727,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 50
      },
      "analysis": {
        "observed": "Native failure after 399 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-28/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-28/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-28/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-28/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-28/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-28/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-28/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-28/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-28/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-28/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-28/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-28/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-28/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-28/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-28/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-28/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-28/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-28/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-28/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-28/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 93,
        "observed_images": 19
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 161,
        "published_frames": 161,
        "decode_ok": true,
        "duration_s": 4.025,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "d0859ce2e0c57dd63bd94a189220c1529e774849be466934a21e8f9f4abea4fe",
        "source_sha256": {
          "third_person": "1c7f52ee42a70bd52bc6d53b6cae0fb9ba05247f6118796ce325295b05cd5f16",
          "left_wrist": "483e4ddf2a5c4473cb38f0c428e6963d388762954cb0c9798e330e8e0316a9cd",
          "right_wrist": "c5d99a5858a09e6f0eb90a8cc01646257fff722cb53cc95180c4eb0c3b43ee56"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "a762ca5d22d4",
          "task": "robotwin/place-a2b-right-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 161,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 161,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-29",
      "variant": "kinex",
      "task_key": "task03/29",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "29",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 19.48,
        "step_limit": 700,
        "steps": 487,
        "success": true,
        "valid": true
      },
      "steps": 487,
      "simulation_time_s": 19.48,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 19.48,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move both the soft bread with golden edges and the soft rectangular bread loaf inside the white basket with beige bottom.",
      "native_instruction": "Move both the soft bread with golden edges and the soft rectangular bread loaf inside the white basket with beige bottom.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9757720936688618,
        "cache_read_input_tokens": 1145856,
        "cache_reported_input_tokens": 1174307,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1174307,
        "cached_input_tokens": 1145856,
        "cost_usd": null,
        "fresh_input_tokens": 28451,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1174307,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1145856,
        "known_fresh_input_tokens": 28451,
        "known_input_tokens": 1174307,
        "known_output_tokens": 5232,
        "known_reasoning_output_tokens": 1056,
        "output_tokens": 5232,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1056,
        "reasoning_reported_output_tokens": 5232,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 43,
          "cache_write_input_tokens": 43,
          "cache_write_reported_input_tokens": 43,
          "cached_input_tokens": 43,
          "fresh_input_tokens": 43,
          "input_tokens": 43,
          "output_tokens": 43,
          "reasoning_output_tokens": 43,
          "reasoning_reported_output_tokens": 43
        },
        "request_attempts": 43,
        "response_count": 43,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1179539,
        "uncached_input_tokens": 28451,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11645,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11685,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11818,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 15676,
            "output_tokens": 244,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 19584,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 20414,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20644,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20835,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21485,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 22064,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22316,
            "output_tokens": 108,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23010,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23281,
            "output_tokens": 313,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24207,
            "output_tokens": 103,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24776,
            "output_tokens": 949,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25747,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26482,
            "output_tokens": 81,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26591,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26905,
            "output_tokens": 82,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27682,
            "output_tokens": 177,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28374,
            "output_tokens": 167,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 29057,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29137,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29329,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29756,
            "output_tokens": 147,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30419,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30611,
            "output_tokens": 347,
            "reasoning_output_tokens": 121
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31393,
            "output_tokens": 440,
            "reasoning_output_tokens": 358
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32134,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32516,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32598,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 32792,
            "output_tokens": 60,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33084,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33476,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33669,
            "output_tokens": 138,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34106,
            "output_tokens": 80,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34485,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34567,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34432,
            "input_tokens": 34760,
            "output_tokens": 77,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 34971,
            "output_tokens": 49,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35082,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35275,
            "output_tokens": 566,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35869,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 49
      },
      "analysis": {
        "observed": "Native success after 487 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-29/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-29/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-29/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-29/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-29/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-29/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-29/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-29/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-29/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-29/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-29/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-29/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-29/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-29/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-29/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-29/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-29/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-29/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-29/workspace/.kinex/tools/cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/aloha-bread-basket.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-29/workspace/.kinex/memos/aloha-bread-basket.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 94,
        "observed_images": 7
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 196,
        "published_frames": 196,
        "decode_ok": true,
        "duration_s": 4.9,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "8dec8e89df2b7dcc3c01c3eec70ff43fdb3a4236ce84571092753041bfdc4e7b",
        "source_sha256": {
          "third_person": "922f6711047278a8eab735a85380ba9c1602b76274e502cc8bc970d58b5321fa",
          "left_wrist": "bec69841923e7f9c87fb3340f63455e5091c63e47185afea582e5aaab23fc38e",
          "right_wrist": "40b499dcd5e8973b84accfe13254b3b97203d85ba3f3321c129b1b1163cb2734"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "18730a4a46fe",
          "task": "robotwin/place-bread-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 196,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 196,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19480000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 196,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 196,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19480000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 196,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 196,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 19480000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-30",
      "variant": "kinex",
      "task_key": "task03/30",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "30",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 14.6,
        "step_limit": 500,
        "steps": 365,
        "success": true,
        "valid": true
      },
      "steps": 365,
      "simulation_time_s": 14.6,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 14.6,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move the spongy golden brown bread from the table into the versatile skillet for frying and sauteing.",
      "native_instruction": "Move the spongy golden brown bread from the table into the versatile skillet for frying and sauteing.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9590967653104377,
        "cache_read_input_tokens": 780160,
        "cache_reported_input_tokens": 813432,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 813432,
        "cached_input_tokens": 780160,
        "cost_usd": null,
        "fresh_input_tokens": 33272,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 813432,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 780160,
        "known_fresh_input_tokens": 33272,
        "known_input_tokens": 813432,
        "known_output_tokens": 4691,
        "known_reasoning_output_tokens": 685,
        "output_tokens": 4691,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 685,
        "reasoning_reported_output_tokens": 4691,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 818123,
        "uncached_input_tokens": 33272,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11646,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11686,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11819,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11869,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 13307,
            "output_tokens": 172,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13312,
            "input_tokens": 16344,
            "output_tokens": 170,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16384,
            "input_tokens": 19812,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 20042,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20232,
            "output_tokens": 224,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 21892,
            "output_tokens": 325,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22696,
            "output_tokens": 68,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23459,
            "output_tokens": 129,
            "reasoning_output_tokens": 88
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24054,
            "output_tokens": 348,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24825,
            "output_tokens": 1111,
            "reasoning_output_tokens": 256
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25959,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26641,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26698,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26826,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27264,
            "output_tokens": 65,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27696,
            "output_tokens": 96,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28097,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28656,
            "output_tokens": 88,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29163,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29515,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30066,
            "output_tokens": 218,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30376,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30817,
            "output_tokens": 65,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31251,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31602,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31821,
            "output_tokens": 57,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31940,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32386,
            "output_tokens": 562,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 32975,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 41
      },
      "analysis": {
        "observed": "Native success after 365 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-30/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-30/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-30/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-30/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-30/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-30/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-30/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-30/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-30/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-30/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-30/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-30/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-30/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-30/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-30/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-30/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-30/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-30/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_step.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-30/workspace/.kinex/tools/cartesian_step.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-30/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 76,
        "observed_images": 12
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 147,
        "published_frames": 147,
        "decode_ok": true,
        "duration_s": 3.675,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "ab7d78611fc68040eba243faaa748c24aba9569948b2ed8a4b1a82e0b29e0517",
        "source_sha256": {
          "third_person": "325e9ae121cb3fb21cd5c5cb04737e265ab0244f7a3c094ce2f606e00902cb9a",
          "left_wrist": "9e48767ce9258bfc50cff091cb6e223662182f72aa08ea8031de69b7b963dd1e",
          "right_wrist": "89b34af05332ffafc9cedad95f38020dce10b46b61df2b52fc8d0e59b35a36f5"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "64600a9a7d9b",
          "task": "robotwin/place-bread-skillet-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 148,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 147,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 148,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 147,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 148,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 147,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 14600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-31",
      "variant": "kinex",
      "task_key": "task03/31",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "31",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 10.4,
        "step_limit": 500,
        "steps": 260,
        "success": true,
        "valid": true
      },
      "steps": 260,
      "simulation_time_s": 10.4,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 10.4,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the compact hamburg box and the Crispy potato fries with red box, place them on the green tray.",
      "native_instruction": "Pick the compact hamburg box and the Crispy potato fries with red box, place them on the green tray.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9622222242823094,
        "cache_read_input_tokens": 1037952,
        "cache_reported_input_tokens": 1078703,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1078703,
        "cached_input_tokens": 1037952,
        "cost_usd": null,
        "fresh_input_tokens": 40751,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1078703,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1037952,
        "known_fresh_input_tokens": 40751,
        "known_input_tokens": 1078703,
        "known_output_tokens": 6117,
        "known_reasoning_output_tokens": 1380,
        "output_tokens": 6117,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1380,
        "reasoning_reported_output_tokens": 6117,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 38,
          "cache_write_input_tokens": 38,
          "cache_write_reported_input_tokens": 38,
          "cached_input_tokens": 38,
          "fresh_input_tokens": 38,
          "input_tokens": 38,
          "output_tokens": 38,
          "reasoning_output_tokens": 38,
          "reasoning_reported_output_tokens": 38
        },
        "request_attempts": 38,
        "response_count": 38,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1084820,
        "uncached_input_tokens": 40751,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11644,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11775,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11823,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 13518,
            "output_tokens": 228,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13568,
            "input_tokens": 17410,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 18658,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 19237,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 19467,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 19658,
            "output_tokens": 219,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 22620,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22720,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23465,
            "output_tokens": 92,
            "reasoning_output_tokens": 53
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24252,
            "output_tokens": 303,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 25556,
            "output_tokens": 974,
            "reasoning_output_tokens": 189
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26551,
            "output_tokens": 115,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 27522,
            "output_tokens": 96,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28348,
            "output_tokens": 101,
            "reasoning_output_tokens": 66
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28511,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 29730,
            "output_tokens": 347,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30992,
            "output_tokens": 119,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31138,
            "output_tokens": 130,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 32034,
            "output_tokens": 377,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32432,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33024,
            "output_tokens": 81,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33310,
            "output_tokens": 112,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33449,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 34184,
            "output_tokens": 80,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34730,
            "output_tokens": 253,
            "reasoning_output_tokens": 93
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35842,
            "output_tokens": 103,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 35967,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36020,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36702,
            "output_tokens": 139,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37704,
            "output_tokens": 240,
            "reasoning_output_tokens": 139
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38591,
            "output_tokens": 215,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39445,
            "output_tokens": 94,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39673,
            "output_tokens": 105,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40063,
            "output_tokens": 850,
            "reasoning_output_tokens": 167
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 40938,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 48
      },
      "analysis": {
        "observed": "Native success after 260 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-31/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-31/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-31/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-31/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-31/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-31/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-31/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-31/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-31/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-31/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-31/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-31/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-31/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-31/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-31/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-31/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-31/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-31/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-31/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/scene_probe.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-31/workspace/.kinex/tools/scene_probe.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-31/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 88,
        "observed_images": 20
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 105,
        "published_frames": 105,
        "decode_ok": true,
        "duration_s": 2.625,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "6d42611f920e56c9201b4f065e757fd98c83c332faa4ace2e9dbe51e753951e2",
        "source_sha256": {
          "third_person": "0fdf13b746c262be4c1ae9986a3fc970a5f1c5a772f473db8a92489440db5b1f",
          "left_wrist": "325fd47c4cc994ca0f11eb465b301d72a042f71414b1fcf97ed08fe06042db09",
          "right_wrist": "5159dd622414d3f76c7f50372cace9d56548b02181e42c403fe7e40803b96871"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "9ab7f3bd3c13",
          "task": "robotwin/place-burger-fries-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 106,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 105,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10400000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 106,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 105,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10400000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 106,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 105,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10400000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-32",
      "variant": "kinex",
      "task_key": "task03/32",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "32",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 25.64,
        "step_limit": 700,
        "steps": 641,
        "success": false,
        "valid": true
      },
      "steps": 641,
      "simulation_time_s": 25.64,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 25.64,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to put the red can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "native_instruction": "Use the left arm to put the red can in the medium basket with smooth texture; the other arm lifts the medium basket with smooth texture.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9696969696969697,
        "cache_read_input_tokens": 2195584,
        "cache_reported_input_tokens": 2264196,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2264196,
        "cached_input_tokens": 2195584,
        "cost_usd": null,
        "fresh_input_tokens": 68612,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2264196,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2195584,
        "known_fresh_input_tokens": 68612,
        "known_input_tokens": 2264196,
        "known_output_tokens": 12105,
        "known_reasoning_output_tokens": 4213,
        "output_tokens": 12105,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 4213,
        "reasoning_reported_output_tokens": 12105,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 56,
          "cache_write_input_tokens": 56,
          "cache_write_reported_input_tokens": 56,
          "cached_input_tokens": 56,
          "fresh_input_tokens": 56,
          "input_tokens": 56,
          "output_tokens": 56,
          "reasoning_output_tokens": 56,
          "reasoning_reported_output_tokens": 56
        },
        "request_attempts": 56,
        "response_count": 56,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2276301,
        "uncached_input_tokens": 68612,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11649,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11689,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11822,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 15955,
            "output_tokens": 276,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 20699,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 21947,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22177,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 23602,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23854,
            "output_tokens": 154,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24363,
            "output_tokens": 322,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25220,
            "output_tokens": 301,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26357,
            "output_tokens": 414,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 29086,
            "output_tokens": 1152,
            "reasoning_output_tokens": 253
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 30259,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 31036,
            "output_tokens": 79,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 33075,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33180,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33997,
            "output_tokens": 403,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 35141,
            "output_tokens": 265,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35607,
            "output_tokens": 138,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 36471,
            "output_tokens": 210,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 37420,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 38079,
            "output_tokens": 129,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38948,
            "output_tokens": 524,
            "reasoning_output_tokens": 390
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38912,
            "input_tokens": 39675,
            "output_tokens": 181,
            "reasoning_output_tokens": 48
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40594,
            "output_tokens": 328,
            "reasoning_output_tokens": 239
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 41590,
            "output_tokens": 56,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41708,
            "output_tokens": 334,
            "reasoning_output_tokens": 231
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42715,
            "output_tokens": 272,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43247,
            "output_tokens": 192,
            "reasoning_output_tokens": 102
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 44109,
            "output_tokens": 117,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43904,
            "input_tokens": 44288,
            "output_tokens": 99,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 45927,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46027,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46388,
            "output_tokens": 452,
            "reasoning_output_tokens": 220
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46208,
            "input_tokens": 47233,
            "output_tokens": 320,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47104,
            "input_tokens": 47946,
            "output_tokens": 259,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 48569,
            "output_tokens": 248,
            "reasoning_output_tokens": 153
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48384,
            "input_tokens": 49422,
            "output_tokens": 145,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49152,
            "input_tokens": 50243,
            "output_tokens": 479,
            "reasoning_output_tokens": 302
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50176,
            "input_tokens": 50913,
            "output_tokens": 170,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50816,
            "input_tokens": 51746,
            "output_tokens": 221,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51584,
            "input_tokens": 52127,
            "output_tokens": 117,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51968,
            "input_tokens": 52912,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52736,
            "input_tokens": 53560,
            "output_tokens": 495,
            "reasoning_output_tokens": 403
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53376,
            "input_tokens": 54726,
            "output_tokens": 136,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54528,
            "input_tokens": 55526,
            "output_tokens": 41,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55296,
            "input_tokens": 55629,
            "output_tokens": 309,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55424,
            "input_tokens": 56128,
            "output_tokens": 296,
            "reasoning_output_tokens": 196
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55936,
            "input_tokens": 57077,
            "output_tokens": 62,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56832,
            "input_tokens": 57201,
            "output_tokens": 225,
            "reasoning_output_tokens": 170
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56960,
            "input_tokens": 58022,
            "output_tokens": 114,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57856,
            "input_tokens": 58742,
            "output_tokens": 253,
            "reasoning_output_tokens": 213
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58624,
            "input_tokens": 59057,
            "output_tokens": 356,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59008,
            "input_tokens": 59440,
            "output_tokens": 611,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59392,
            "input_tokens": 60076,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 69
      },
      "analysis": {
        "observed": "Native failure after 641 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-32/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-32/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-32/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-32/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-32/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-32/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-32/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-32/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-32/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-32/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-32/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-32/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-32/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-32/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-32/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-32/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-32/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-32/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_step.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-32/workspace/.kinex/tools/cartesian_step.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-can-basket.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-32/workspace/.kinex/memos/robotwin-can-basket.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 127,
        "observed_images": 21
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 258,
        "published_frames": 258,
        "decode_ok": true,
        "duration_s": 6.45,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "fc2f54c763430d2b8a4074f909637fe4f2b7c93a024df50ce71d02387aeb3b4a",
        "source_sha256": {
          "third_person": "a91494051fcee894acaafc83fc9697b269dec607fe14bb6aa6d1069f40d67193",
          "left_wrist": "d3aa3e32a4fb495b9621e3cfc846efa870a01edd6db1ea39832d493e2ccb6683",
          "right_wrist": "e273171d5d5e239ed82d4efbcec0bb5ce57ad6b5d9b00f0a0da3a21767f06028"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "43fe2bf1cae3",
          "task": "robotwin/place-can-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 258,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 258,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 25640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 258,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 258,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 25640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 258,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 258,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 25640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-33",
      "variant": "kinex",
      "task_key": "task03/33",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "33",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 15.88,
        "step_limit": 800,
        "steps": 397,
        "success": true,
        "valid": true
      },
      "steps": 397,
      "simulation_time_s": 15.88,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 15.88,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the small red soda can and the lightweight cylindrical metal can in parallel, then put both into the rectangular plasticbox.",
      "native_instruction": "Grab the small red soda can and the lightweight cylindrical metal can in parallel, then put both into the rectangular plasticbox.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9723858054935874,
        "cache_read_input_tokens": 1130240,
        "cache_reported_input_tokens": 1162337,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1162337,
        "cached_input_tokens": 1130240,
        "cost_usd": null,
        "fresh_input_tokens": 32097,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1162337,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1130240,
        "known_fresh_input_tokens": 32097,
        "known_input_tokens": 1162337,
        "known_output_tokens": 6227,
        "known_reasoning_output_tokens": 1423,
        "output_tokens": 6227,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1423,
        "reasoning_reported_output_tokens": 6227,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 42,
          "cache_write_input_tokens": 42,
          "cache_write_reported_input_tokens": 42,
          "cached_input_tokens": 42,
          "fresh_input_tokens": 42,
          "input_tokens": 42,
          "output_tokens": 42,
          "reasoning_output_tokens": 42,
          "reasoning_reported_output_tokens": 42
        },
        "request_attempts": 42,
        "response_count": 42,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1168564,
        "uncached_input_tokens": 32097,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11646,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11694,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11825,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11908,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12059,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12194,
            "output_tokens": 167,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 15882,
            "output_tokens": 144,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 18896,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18816,
            "input_tokens": 20143,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20096,
            "input_tokens": 20373,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 20566,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 20818,
            "output_tokens": 175,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21388,
            "output_tokens": 344,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 23349,
            "output_tokens": 214,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24722,
            "output_tokens": 1258,
            "reasoning_output_tokens": 145
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26002,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26703,
            "output_tokens": 103,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26834,
            "output_tokens": 76,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27212,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27801,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 27891,
            "output_tokens": 73,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 27992,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28584,
            "output_tokens": 173,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29502,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30041,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30859,
            "output_tokens": 416,
            "reasoning_output_tokens": 199
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31734,
            "output_tokens": 371,
            "reasoning_output_tokens": 288
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32798,
            "output_tokens": 201,
            "reasoning_output_tokens": 120
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32896,
            "input_tokens": 33695,
            "output_tokens": 58,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34932,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35722,
            "output_tokens": 119,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 35869,
            "output_tokens": 79,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36688,
            "output_tokens": 127,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37294,
            "output_tokens": 91,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37865,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38265,
            "output_tokens": 189,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 39200,
            "output_tokens": 93,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39767,
            "output_tokens": 101,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40002,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40093,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40401,
            "output_tokens": 702,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41128,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 49
      },
      "analysis": {
        "observed": "Native success after 397 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-33/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-33/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-33/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-33/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-33/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-33/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-33/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-33/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-33/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-33/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-33/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-33/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-33/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-33/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-33/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-33/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-33/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-33/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/dual_cartesian.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-33/workspace/.kinex/tools/dual_cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/aloha_can_placement.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-33/workspace/.kinex/memos/aloha_can_placement.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 93,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 160,
        "published_frames": 160,
        "decode_ok": true,
        "duration_s": 4.0,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "e3ffd3f4e192174dfce9ada45daf902269da5ee6877b0625709fa3163a5981c8",
        "source_sha256": {
          "third_person": "32b2bec8d4072ec78ab6b338d838d6eba8ea9b6ba175281923c4763729509dd8",
          "left_wrist": "05370e2103a9304a6e727ca5701238e47d149b4332c02f4ce02cec1f53feb830",
          "right_wrist": "5984e836f68589cc7d72402e523e9ae412fd0b77c2a9ade3d6bb7ad9fc715946"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "d748d21ad4b0",
          "task": "robotwin/place-cans-plasticbox-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 160,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 160,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15880000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 160,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 160,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15880000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 160,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 160,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 15880000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-34",
      "variant": "kinex",
      "task_key": "task03/34",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "34",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.0,
        "step_limit": 400,
        "steps": 225,
        "success": true,
        "valid": true
      },
      "steps": 225,
      "simulation_time_s": 9.0,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 9.0,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the ceramic drinking cup with floral design on the gray plate for dining.",
      "native_instruction": "Place the ceramic drinking cup with floral design on the gray plate for dining.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9220818268257803,
        "cache_read_input_tokens": 604160,
        "cache_reported_input_tokens": 655213,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 655213,
        "cached_input_tokens": 604160,
        "cost_usd": null,
        "fresh_input_tokens": 51053,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 655213,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 604160,
        "known_fresh_input_tokens": 51053,
        "known_input_tokens": 655213,
        "known_output_tokens": 4293,
        "known_reasoning_output_tokens": 522,
        "output_tokens": 4293,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 522,
        "reasoning_reported_output_tokens": 4293,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 30,
          "cache_write_input_tokens": 30,
          "cache_write_reported_input_tokens": 30,
          "cached_input_tokens": 30,
          "fresh_input_tokens": 30,
          "input_tokens": 30,
          "output_tokens": 30,
          "reasoning_output_tokens": 30,
          "reasoning_reported_output_tokens": 30
        },
        "request_attempts": 30,
        "response_count": 30,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 659506,
        "uncached_input_tokens": 51053,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11637,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11685,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11816,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11874,
            "output_tokens": 21,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11909,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11945,
            "output_tokens": 223,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 16600,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 17847,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18426,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 18656,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 18847,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 19099,
            "output_tokens": 187,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 19715,
            "output_tokens": 178,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 21250,
            "output_tokens": 138,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 22547,
            "output_tokens": 248,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23519,
            "output_tokens": 494,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24364,
            "output_tokens": 1055,
            "reasoning_output_tokens": 196
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25440,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26157,
            "output_tokens": 80,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26264,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 26726,
            "output_tokens": 88,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 27190,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27600,
            "output_tokens": 87,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28063,
            "output_tokens": 179,
            "reasoning_output_tokens": 91
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 28617,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 29077,
            "output_tokens": 77,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29286,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29380,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29571,
            "output_tokens": 512,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30106,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 36
      },
      "analysis": {
        "observed": "Native success after 225 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-34/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-34/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-34/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-34/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-34/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-34/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-34/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-34/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-34/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-34/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-34/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-34/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-34/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-34/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-34/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-34/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-34/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-34/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-34/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/aloha-placement.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-34/workspace/.kinex/memos/aloha-placement.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 68,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 91,
        "published_frames": 91,
        "decode_ok": true,
        "duration_s": 2.275,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "a31b346b119e7840f0231061dc589b17867b7864a17e44b9f8ec89ce36c93adc",
        "source_sha256": {
          "third_person": "5a078e907ff257762bd2600ad616ab6f0d8620de7a8603752b704babdc04a492",
          "left_wrist": "02a0b85fad43fab7b39aa44a20a67c44bf27acab2a5c46d32086037d22c77b96",
          "right_wrist": "a9b6a78d596447350b5576199f75e9f7f729b126a7828e06fd5ee23c3f8ef22a"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "b5facf7d5107",
          "task": "robotwin/place-container-plate-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 92,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 91,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9000000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 92,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 91,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9000000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 92,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 91,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9000000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-35",
      "variant": "kinex",
      "task_key": "task03/35",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "35",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 13.36,
        "step_limit": 500,
        "steps": 334,
        "success": true,
        "valid": true
      },
      "steps": 334,
      "simulation_time_s": 13.36,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 13.36,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Place the small smooth cup onto the light gray coaster gently.",
      "native_instruction": "Place the small smooth cup onto the light gray coaster gently.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9700975852980982,
        "cache_read_input_tokens": 953344,
        "cache_reported_input_tokens": 982730,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 982730,
        "cached_input_tokens": 953344,
        "cost_usd": null,
        "fresh_input_tokens": 29386,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 982730,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 953344,
        "known_fresh_input_tokens": 29386,
        "known_input_tokens": 982730,
        "known_output_tokens": 5520,
        "known_reasoning_output_tokens": 924,
        "output_tokens": 5520,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 924,
        "reasoning_reported_output_tokens": 5520,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 36,
          "cache_write_input_tokens": 36,
          "cache_write_reported_input_tokens": 36,
          "cached_input_tokens": 36,
          "fresh_input_tokens": 36,
          "input_tokens": 36,
          "output_tokens": 36,
          "reasoning_output_tokens": 36,
          "reasoning_reported_output_tokens": 36
        },
        "request_attempts": 36,
        "response_count": 36,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 988250,
        "uncached_input_tokens": 29386,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11635,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11683,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11814,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 13152,
            "output_tokens": 215,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13184,
            "input_tokens": 16268,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16128,
            "input_tokens": 18784,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 19363,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 19527,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 19717,
            "output_tokens": 148,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 20573,
            "output_tokens": 111,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 23022,
            "output_tokens": 59,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23313,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23447,
            "output_tokens": 323,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 25032,
            "output_tokens": 400,
            "reasoning_output_tokens": 76
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 26481,
            "output_tokens": 313,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27018,
            "output_tokens": 125,
            "reasoning_output_tokens": 83
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 28322,
            "output_tokens": 1002,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29345,
            "output_tokens": 79,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29472,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 30177,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30256,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30836,
            "output_tokens": 139,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31002,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31710,
            "output_tokens": 119,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32413,
            "output_tokens": 138,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32384,
            "input_tokens": 33148,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33789,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34468,
            "output_tokens": 241,
            "reasoning_output_tokens": 156
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35302,
            "output_tokens": 380,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35893,
            "output_tokens": 125,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36608,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36608,
            "input_tokens": 37293,
            "output_tokens": 70,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37497,
            "output_tokens": 59,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37618,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38068,
            "output_tokens": 591,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38684,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 46
      },
      "analysis": {
        "observed": "Native success after 334 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-35/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-35/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-35/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-35/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-35/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-35/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-35/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-35/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-35/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-35/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-35/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-35/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-35/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-35/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-35/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-35/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-35/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-35/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-35/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-cup-placement.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-35/workspace/.kinex/memos/robotwin-cup-placement.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 84,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 135,
        "published_frames": 135,
        "decode_ok": true,
        "duration_s": 3.375,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "5d3085df7f2290b2f59c6246f98e89d5438c66097be16f7779ac2b956b668c13",
        "source_sha256": {
          "third_person": "0ef2de3767b8acf08932827119dcee9cc5207651de6fcff2dfc41b1db8eb3313",
          "left_wrist": "25c1d64c90c57afcaa0f3a55d02a5482f199c4b8ab1ae40d70af2550baa8eb90",
          "right_wrist": "4a0488b20c994d6cac2ddd179ea81a38c23711344006bc9f0b5087de244d338d"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "4bb2608d375f",
          "task": "robotwin/place-empty-cup-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 135,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 135,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13360000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 135,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 135,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13360000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 135,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 135,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13360000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-36",
      "variant": "kinex",
      "task_key": "task03/36",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "36",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 8.96,
        "step_limit": 400,
        "steps": 224,
        "success": false,
        "valid": true
      },
      "steps": 224,
      "simulation_time_s": 8.96,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 8.96,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick the beige circular blade fan, place it on the Pink mat, face it toward the robot",
      "native_instruction": "Pick the beige circular blade fan, place it on the Pink mat, face it toward the robot",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9719530730098324,
        "cache_read_input_tokens": 1142144,
        "cache_reported_input_tokens": 1175102,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1175102,
        "cached_input_tokens": 1142144,
        "cost_usd": null,
        "fresh_input_tokens": 32958,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1175102,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1142144,
        "known_fresh_input_tokens": 32958,
        "known_input_tokens": 1175102,
        "known_output_tokens": 7119,
        "known_reasoning_output_tokens": 1762,
        "output_tokens": 7119,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1762,
        "reasoning_reported_output_tokens": 7119,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 38,
          "cache_write_input_tokens": 38,
          "cache_write_reported_input_tokens": 38,
          "cached_input_tokens": 38,
          "fresh_input_tokens": 38,
          "input_tokens": 38,
          "output_tokens": 38,
          "reasoning_output_tokens": 38,
          "reasoning_reported_output_tokens": 38
        },
        "request_attempts": 38,
        "response_count": 38,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1182221,
        "uncached_input_tokens": 32958,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11640,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11680,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11813,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11861,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 14686,
            "output_tokens": 192,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14720,
            "input_tokens": 18521,
            "output_tokens": 209,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 22788,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23082,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23271,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23523,
            "output_tokens": 228,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 25160,
            "output_tokens": 215,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25670,
            "output_tokens": 333,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 27282,
            "output_tokens": 240,
            "reasoning_output_tokens": 115
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 29858,
            "output_tokens": 302,
            "reasoning_output_tokens": 116
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30718,
            "output_tokens": 964,
            "reasoning_output_tokens": 212
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31703,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 31860,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32140,
            "output_tokens": 218,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32128,
            "input_tokens": 32691,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33168,
            "output_tokens": 103,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33684,
            "output_tokens": 196,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34295,
            "output_tokens": 108,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34818,
            "output_tokens": 51,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 34891,
            "output_tokens": 167,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35479,
            "output_tokens": 306,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 37267,
            "output_tokens": 90,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37680,
            "output_tokens": 197,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38295,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38487,
            "output_tokens": 317,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 40017,
            "output_tokens": 452,
            "reasoning_output_tokens": 359
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40893,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 41086,
            "output_tokens": 349,
            "reasoning_output_tokens": 256
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41611,
            "output_tokens": 188,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41728,
            "input_tokens": 42215,
            "output_tokens": 45,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42322,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 42515,
            "output_tokens": 416,
            "reasoning_output_tokens": 131
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42752,
            "input_tokens": 42958,
            "output_tokens": 491,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43474,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 48
      },
      "analysis": {
        "observed": "Native failure after 224 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-36/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-36/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-36/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-36/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-36/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-36/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-36/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-36/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-36/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-36/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-36/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-36/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-36/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-36/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-36/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-36/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-36/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-36/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-36/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-36/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 88,
        "observed_images": 16
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 91,
        "published_frames": 91,
        "decode_ok": true,
        "duration_s": 2.275,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "04722ad564f48ebec5d897bd22ba6763969acc306c1081b110d5ff3f1262e7f4",
        "source_sha256": {
          "third_person": "1041ecbc91d3e786324060dda049917465c523be9aaed377bcce71ee0fd218fa",
          "left_wrist": "5df94dc24353ac4cfa8af45b491c1064ca7cb9c23153b891832e6396a6643648",
          "right_wrist": "7b342c4654d266cd0e21935216ef7bf6724812453899f7b47d0a5eb981c73564"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "9e82d0551ee9",
          "task": "robotwin/place-fan-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 91,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 91,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8960000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 91,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 91,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8960000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 91,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 91,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 8960000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-37",
      "variant": "kinex",
      "task_key": "task03/37",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "37",
      "seed": 2,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 2
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.040000000000001,
        "step_limit": 400,
        "steps": 226,
        "success": true,
        "valid": true
      },
      "steps": 226,
      "simulation_time_s": 9.040000000000001,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 9.040000000000001,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the dark gray computer mouse with logo with the left arm and drop it onto the Cyan mat.",
      "native_instruction": "Pick up the dark gray computer mouse with logo with the left arm and drop it onto the Cyan mat.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9441454700374713,
        "cache_read_input_tokens": 487552,
        "cache_reported_input_tokens": 516395,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 516395,
        "cached_input_tokens": 487552,
        "cost_usd": null,
        "fresh_input_tokens": 28843,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 516395,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 487552,
        "known_fresh_input_tokens": 28843,
        "known_input_tokens": 516395,
        "known_output_tokens": 3586,
        "known_reasoning_output_tokens": 533,
        "output_tokens": 3586,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 533,
        "reasoning_reported_output_tokens": 3586,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 24,
          "cache_write_input_tokens": 24,
          "cache_write_reported_input_tokens": 24,
          "cached_input_tokens": 24,
          "fresh_input_tokens": 24,
          "input_tokens": 24,
          "output_tokens": 24,
          "reasoning_output_tokens": 24,
          "reasoning_reported_output_tokens": 24
        },
        "request_attempts": 24,
        "response_count": 24,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 519981,
        "uncached_input_tokens": 28843,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11644,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11684,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11817,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 13823,
            "output_tokens": 199,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13824,
            "input_tokens": 16904,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16768,
            "input_tokens": 18151,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18315,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 19594,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 19846,
            "output_tokens": 86,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 20164,
            "output_tokens": 300,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 20999,
            "output_tokens": 296,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 22376,
            "output_tokens": 1039,
            "reasoning_output_tokens": 223
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23436,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 23695,
            "output_tokens": 63,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24266,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24829,
            "output_tokens": 81,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24704,
            "input_tokens": 24932,
            "output_tokens": 95,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25527,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26097,
            "output_tokens": 213,
            "reasoning_output_tokens": 141
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26812,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27375,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27567,
            "output_tokens": 65,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28034,
            "output_tokens": 449,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28508,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 31
      },
      "analysis": {
        "observed": "Native success after 226 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-37/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-37/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-37/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-37/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-37/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-37/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-37/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-37/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-37/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-37/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-37/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-37/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-37/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-37/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-37/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-37/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-37/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-37/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-37/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-37/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 57,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 2,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 92,
        "published_frames": 92,
        "decode_ok": true,
        "duration_s": 2.3,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "4f2660206af8689c355b10ae24e309a9d64e2afc4f847724779ce8c97fbe16c8",
        "source_sha256": {
          "third_person": "af6786c45b3a54562a61af29164ded60945a631bad77e4a9c4e176d0cd82dac2",
          "left_wrist": "37462e2516ad571dcfd4f6608275913dbc7d62b6bd5894ec4d31a7ffd8adbf01",
          "right_wrist": "fee5082165e02d5175f0235c74316ce1eb3ed65cf3413bfb0474df9a0682155d"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "b129fa3e4dab",
          "task": "robotwin/place-mouse-pad-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 92,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 92,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9040000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 92,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 92,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9040000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 92,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 92,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9040000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-38",
      "variant": "kinex",
      "task_key": "task03/38",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "38",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 17.88,
        "step_limit": 700,
        "steps": 447,
        "success": false,
        "valid": true
      },
      "steps": 447,
      "simulation_time_s": 17.88,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 17.88,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the toycar with rounded edges and place into the basket with rounded edges.",
      "native_instruction": "Grab the toycar with rounded edges and place into the basket with rounded edges.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9666864154053401,
        "cache_read_input_tokens": 1168256,
        "cache_reported_input_tokens": 1208516,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1208516,
        "cached_input_tokens": 1168256,
        "cost_usd": null,
        "fresh_input_tokens": 40260,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1208516,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1168256,
        "known_fresh_input_tokens": 40260,
        "known_input_tokens": 1208516,
        "known_output_tokens": 7739,
        "known_reasoning_output_tokens": 1658,
        "output_tokens": 7739,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1658,
        "reasoning_reported_output_tokens": 7739,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 42,
          "cache_write_input_tokens": 42,
          "cache_write_reported_input_tokens": 42,
          "cached_input_tokens": 42,
          "fresh_input_tokens": 42,
          "input_tokens": 42,
          "output_tokens": 42,
          "reasoning_output_tokens": 42,
          "reasoning_reported_output_tokens": 42
        },
        "request_attempts": 42,
        "response_count": 42,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1216255,
        "uncached_input_tokens": 40260,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11637,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11677,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11810,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11860,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 13634,
            "output_tokens": 211,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13696,
            "input_tokens": 16734,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 19250,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 19414,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 19605,
            "output_tokens": 68,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 20213,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20096,
            "input_tokens": 20465,
            "output_tokens": 259,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 21119,
            "output_tokens": 259,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 22322,
            "output_tokens": 165,
            "reasoning_output_tokens": 7
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 25213,
            "output_tokens": 179,
            "reasoning_output_tokens": 139
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25456,
            "output_tokens": 292,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26298,
            "output_tokens": 309,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 27215,
            "output_tokens": 288,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28038,
            "output_tokens": 278,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 29119,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 30082,
            "output_tokens": 711,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30816,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31092,
            "output_tokens": 306,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31772,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32195,
            "output_tokens": 226,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32742,
            "output_tokens": 82,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33152,
            "output_tokens": 178,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33660,
            "output_tokens": 285,
            "reasoning_output_tokens": 173
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34174,
            "output_tokens": 121,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34624,
            "output_tokens": 103,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34749,
            "output_tokens": 254,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 35333,
            "output_tokens": 376,
            "reasoning_output_tokens": 261
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36425,
            "output_tokens": 126,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37267,
            "output_tokens": 125,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37566,
            "output_tokens": 150,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38431,
            "output_tokens": 506,
            "reasoning_output_tokens": 248
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39095,
            "output_tokens": 189,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39613,
            "output_tokens": 156,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39424,
            "input_tokens": 40097,
            "output_tokens": 108,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40533,
            "output_tokens": 114,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40975,
            "output_tokens": 220,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41224,
            "output_tokens": 570,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 41820,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 50
      },
      "analysis": {
        "observed": "Native failure after 447 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-38/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-38/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-38/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-38/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-38/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-38/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-38/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-38/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-38/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-38/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-38/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-38/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-38/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-38/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-38/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-38/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-38/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-38/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-38/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/aloha_basket.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-38/workspace/.kinex/memos/aloha_basket.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 94,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 180,
        "published_frames": 180,
        "decode_ok": true,
        "duration_s": 4.5,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "1eb120203d3e176dc5c1c9078dffdc73a64806b11aaedad771642939b0c239ed",
        "source_sha256": {
          "third_person": "142e3a8ac80b64ba3f873f89619bc907b7c4ce718badc5f7837eefcbdc394647",
          "left_wrist": "aeee598cb4610f7fe3ccd5c0f7f8197e57e1513da28eac14143899b8f11d9726",
          "right_wrist": "b1c4920c592d57777b8df93535f0421c0456fa65a913fc56ab2ed49bad9ad166"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "f85550e95dc8",
          "task": "robotwin/place-object-basket-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 180,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 180,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17880000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 180,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 180,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17880000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 180,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 180,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17880000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-39",
      "variant": "kinex",
      "task_key": "task03/39",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "39",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.28,
        "step_limit": 400,
        "steps": 232,
        "success": true,
        "valid": true
      },
      "steps": 232,
      "simulation_time_s": 9.28,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 9.28,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use one arm to pick the palm-sized black stapler for the black screen electronicscale",
      "native_instruction": "Use one arm to pick the palm-sized black stapler for the black screen electronicscale",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9630797468056569,
        "cache_read_input_tokens": 654848,
        "cache_reported_input_tokens": 679952,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 679952,
        "cached_input_tokens": 654848,
        "cost_usd": null,
        "fresh_input_tokens": 25104,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 679952,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 654848,
        "known_fresh_input_tokens": 25104,
        "known_input_tokens": 679952,
        "known_output_tokens": 4056,
        "known_reasoning_output_tokens": 444,
        "output_tokens": 4056,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 444,
        "reasoning_reported_output_tokens": 4056,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 31,
          "cache_write_input_tokens": 31,
          "cache_write_reported_input_tokens": 31,
          "cached_input_tokens": 31,
          "fresh_input_tokens": 31,
          "input_tokens": 31,
          "output_tokens": 31,
          "reasoning_output_tokens": 31,
          "reasoning_reported_output_tokens": 31
        },
        "request_attempts": 31,
        "response_count": 31,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 684008,
        "uncached_input_tokens": 25104,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11637,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11768,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11816,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 12064,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12511,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 15905,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 17153,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17024,
            "input_tokens": 17317,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17152,
            "input_tokens": 17510,
            "output_tokens": 180,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 20176,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20428,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 20530,
            "output_tokens": 296,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 21221,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21492,
            "output_tokens": 291,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21376,
            "input_tokens": 23056,
            "output_tokens": 853,
            "reasoning_output_tokens": 150
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23930,
            "output_tokens": 140,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23808,
            "input_tokens": 24247,
            "output_tokens": 99,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24958,
            "output_tokens": 52,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25032,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25113,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24960,
            "input_tokens": 25603,
            "output_tokens": 109,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25846,
            "output_tokens": 149,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26363,
            "output_tokens": 106,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26491,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26904,
            "output_tokens": 95,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27363,
            "output_tokens": 195,
            "reasoning_output_tokens": 97
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27916,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28377,
            "output_tokens": 68,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28579,
            "output_tokens": 71,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28416,
            "input_tokens": 29050,
            "output_tokens": 521,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29596,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 37
      },
      "analysis": {
        "observed": "Native success after 232 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-39/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-39/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-39/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-39/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-39/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-39/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-39/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-39/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-39/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-39/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-39/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-39/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-39/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-39/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-39/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-39/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-39/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-39/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-39/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-arx5.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-39/workspace/.kinex/memos/robotwin-arx5.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 70,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 94,
        "published_frames": 94,
        "decode_ok": true,
        "duration_s": 2.35,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "24bf8b06a7170c58d7811829a77547c2f032361f4d1c2a2ea228598d0baf9a98",
        "source_sha256": {
          "third_person": "be2144bd23c6ae8afc93ae85b922d049edfa08d922b7d16b3ea3fcd7cfebc438",
          "left_wrist": "5ed6165d42139c4c3674f537b888e0189c91e6479e57fefbb9d7b568efc28f6a",
          "right_wrist": "1bec73ea14872554e03c9749961a222fa31d0b37ff1a4966d814dd21764628e9"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ad1f11519f0e",
          "task": "robotwin/place-object-scale-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 94,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 94,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9280000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 94,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 94,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9280000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 94,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 94,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9280000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-40",
      "variant": "kinex",
      "task_key": "task03/40",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "40",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.8,
        "step_limit": 400,
        "steps": 245,
        "success": true,
        "valid": true
      },
      "steps": 245,
      "simulation_time_s": 9.8,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 9.8,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set the rectangular remotecontrol on the displaystand with hollow center.",
      "native_instruction": "Set the rectangular remotecontrol on the displaystand with hollow center.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9570301614036805,
        "cache_read_input_tokens": 704000,
        "cache_reported_input_tokens": 735609,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 735609,
        "cached_input_tokens": 704000,
        "cost_usd": null,
        "fresh_input_tokens": 31609,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 735609,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 704000,
        "known_fresh_input_tokens": 31609,
        "known_input_tokens": 735609,
        "known_output_tokens": 4905,
        "known_reasoning_output_tokens": 1131,
        "output_tokens": 4905,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1131,
        "reasoning_reported_output_tokens": 4905,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 32,
          "cache_write_input_tokens": 32,
          "cache_write_reported_input_tokens": 32,
          "cached_input_tokens": 32,
          "fresh_input_tokens": 32,
          "input_tokens": 32,
          "output_tokens": 32,
          "reasoning_output_tokens": 32,
          "reasoning_reported_output_tokens": 32
        },
        "request_attempts": 32,
        "response_count": 32,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 740514,
        "uncached_input_tokens": 31609,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11634,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11767,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11815,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12063,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12275,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 16108,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 17356,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 17586,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 17777,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 18356,
            "output_tokens": 122,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 19917,
            "output_tokens": 179,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20797,
            "output_tokens": 284,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 22096,
            "output_tokens": 332,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 22793,
            "output_tokens": 134,
            "reasoning_output_tokens": 95
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 23622,
            "output_tokens": 740,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 24383,
            "output_tokens": 45,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 24524,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25253,
            "output_tokens": 160,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25440,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26110,
            "output_tokens": 124,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26809,
            "output_tokens": 89,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 26920,
            "output_tokens": 79,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27523,
            "output_tokens": 212,
            "reasoning_output_tokens": 116
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28314,
            "output_tokens": 430,
            "reasoning_output_tokens": 218
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 28989,
            "output_tokens": 264,
            "reasoning_output_tokens": 166
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29321,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29992,
            "output_tokens": 134,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30704,
            "output_tokens": 155,
            "reasoning_output_tokens": 110
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 30993,
            "output_tokens": 54,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31109,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31378,
            "output_tokens": 484,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31885,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 38
      },
      "analysis": {
        "observed": "Native success after 245 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-40/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-40/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-40/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-40/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-40/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-40/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-40/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-40/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-40/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-40/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-40/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-40/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-40/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-40/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-40/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-40/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-40/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-40/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-40/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-placement.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-40/workspace/.kinex/memos/robotwin-placement.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 72,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 99,
        "published_frames": 99,
        "decode_ok": true,
        "duration_s": 2.475,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "0261731aa5a8fad6bca958350edaaf19dd23fc25375ddcd69da37893e29c5efa",
        "source_sha256": {
          "third_person": "f07e3d4533a8aab994bcafe586d4148f8bfac2722f1fd5ea34e9bf55df5b1fb4",
          "left_wrist": "6851a6db0263c87fe9cc298c15aa31015a6192df58d960f32b0deddf182b94ae",
          "right_wrist": "ac70ae5aa80e7e1970b840ef5596c4aa6eec207a85b62193373efcba8e052a6a"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "6a0604550308",
          "task": "robotwin/place-object-stand-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 100,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 99,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9800000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 100,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 99,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9800000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 100,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 99,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9800000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-41",
      "variant": "kinex",
      "task_key": "task03/41",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "41",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.64,
        "step_limit": 400,
        "steps": 241,
        "success": true,
        "valid": true
      },
      "steps": 241,
      "simulation_time_s": 9.64,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 9.64,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Set the phone with shiny black back on the L-shaped smooth brown phonestand.",
      "native_instruction": "Set the phone with shiny black back on the L-shaped smooth brown phonestand.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9607177568743079,
        "cache_read_input_tokens": 851072,
        "cache_reported_input_tokens": 885871,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 885871,
        "cached_input_tokens": 851072,
        "cost_usd": null,
        "fresh_input_tokens": 34799,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 885871,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 851072,
        "known_fresh_input_tokens": 34799,
        "known_input_tokens": 885871,
        "known_output_tokens": 7242,
        "known_reasoning_output_tokens": 2264,
        "output_tokens": 7242,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2264,
        "reasoning_reported_output_tokens": 7242,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 34,
          "cache_write_input_tokens": 34,
          "cache_write_reported_input_tokens": 34,
          "cached_input_tokens": 34,
          "fresh_input_tokens": 34,
          "input_tokens": 34,
          "output_tokens": 34,
          "reasoning_output_tokens": 34,
          "reasoning_reported_output_tokens": 34
        },
        "request_attempts": 34,
        "response_count": 34,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 893113,
        "uncached_input_tokens": 34799,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11636,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11684,
            "output_tokens": 146,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 12137,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12273,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 16783,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16640,
            "input_tokens": 18030,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18260,
            "output_tokens": 141,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 20264,
            "output_tokens": 89,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 20563,
            "output_tokens": 267,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 21225,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21120,
            "input_tokens": 21496,
            "output_tokens": 260,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 22976,
            "output_tokens": 437,
            "reasoning_output_tokens": 120
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23296,
            "input_tokens": 24064,
            "output_tokens": 1287,
            "reasoning_output_tokens": 431
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25371,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25533,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 26287,
            "output_tokens": 125,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 26438,
            "output_tokens": 92,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 26828,
            "output_tokens": 117,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27240,
            "output_tokens": 78,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27619,
            "output_tokens": 164,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28088,
            "output_tokens": 834,
            "reasoning_output_tokens": 471
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 30277,
            "output_tokens": 300,
            "reasoning_output_tokens": 193
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 31473,
            "output_tokens": 211,
            "reasoning_output_tokens": 121
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31985,
            "output_tokens": 80,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32836,
            "output_tokens": 275,
            "reasoning_output_tokens": 183
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33024,
            "input_tokens": 33415,
            "output_tokens": 83,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33516,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33709,
            "output_tokens": 171,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34336,
            "output_tokens": 295,
            "reasoning_output_tokens": 54
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34985,
            "output_tokens": 219,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35508,
            "output_tokens": 149,
            "reasoning_output_tokens": 103
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35791,
            "output_tokens": 102,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36288,
            "output_tokens": 646,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 36957,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 46
      },
      "analysis": {
        "observed": "Native success after 241 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-41/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-41/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-41/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-41/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-41/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-41/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-41/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-41/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-41/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-41/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-41/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-41/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-41/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-41/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-41/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-41/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-41/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-41/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-41/workspace/.kinex/tools/cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/phone-stand.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-41/workspace/.kinex/memos/phone-stand.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 82,
        "observed_images": 13
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 98,
        "published_frames": 98,
        "decode_ok": true,
        "duration_s": 2.45,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "70e707d4cb481e4bb460c4a269a5ab5aa822ee18a5f4a18b2f04f9a442ae86d8",
        "source_sha256": {
          "third_person": "16705757be2fbc088680087e76a2d56a7dd4b9eef979ffd0ef8e0cc3d18491c3",
          "left_wrist": "39436cd010b6b8e6058d64837f5b7af8185f974ee8e78c9f153aa322e7601c45",
          "right_wrist": "75388f3010717248af67754837d8a136a6d2453c43606b0e4cabc0906cfaabce"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "57e645033ebb",
          "task": "robotwin/place-phone-stand-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 98,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9640000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 98,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9640000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 98,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9640000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-42",
      "variant": "kinex",
      "task_key": "task03/42",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "42",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.6,
        "step_limit": 500,
        "steps": 240,
        "success": true,
        "valid": true
      },
      "steps": 240,
      "simulation_time_s": 9.6,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 9.6,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the casual brown shoe off the table and drop it on the mat",
      "native_instruction": "Grab the casual brown shoe off the table and drop it on the mat",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9666522664077589,
        "cache_read_input_tokens": 651136,
        "cache_reported_input_tokens": 673599,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 673599,
        "cached_input_tokens": 651136,
        "cost_usd": null,
        "fresh_input_tokens": 22463,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 673599,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 651136,
        "known_fresh_input_tokens": 22463,
        "known_input_tokens": 673599,
        "known_output_tokens": 4186,
        "known_reasoning_output_tokens": 624,
        "output_tokens": 4186,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 624,
        "reasoning_reported_output_tokens": 4186,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 29,
          "cache_write_input_tokens": 29,
          "cache_write_reported_input_tokens": 29,
          "cached_input_tokens": 29,
          "fresh_input_tokens": 29,
          "input_tokens": 29,
          "output_tokens": 29,
          "reasoning_output_tokens": 29,
          "reasoning_reported_output_tokens": 29
        },
        "request_attempts": 29,
        "response_count": 29,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 677785,
        "uncached_input_tokens": 22463,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11637,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11677,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11810,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 14993,
            "output_tokens": 117,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14976,
            "input_tokens": 15668,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15616,
            "input_tokens": 15884,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 15944,
            "output_tokens": 146,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16000,
            "input_tokens": 19201,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 19431,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 19623,
            "output_tokens": 58,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20221,
            "output_tokens": 180,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 22428,
            "output_tokens": 267,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 23090,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 23361,
            "output_tokens": 268,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24164,
            "output_tokens": 265,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 25012,
            "output_tokens": 271,
            "reasoning_output_tokens": 229
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26462,
            "output_tokens": 729,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27214,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27380,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27830,
            "output_tokens": 164,
            "reasoning_output_tokens": 29
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28356,
            "output_tokens": 66,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28786,
            "output_tokens": 91,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29243,
            "output_tokens": 350,
            "reasoning_output_tokens": 225
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29440,
            "input_tokens": 29951,
            "output_tokens": 115,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29952,
            "input_tokens": 30428,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30610,
            "output_tokens": 50,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30722,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30990,
            "output_tokens": 466,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 31483,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 37
      },
      "analysis": {
        "observed": "Native success after 240 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-42/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-42/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-42/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-42/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-42/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-42/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-42/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-42/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-42/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-42/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-42/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-42/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-42/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-42/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-42/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-42/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-42/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-42/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-42/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-manipulation.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-42/workspace/.kinex/memos/robotwin-manipulation.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 68,
        "observed_images": 8
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 97,
        "published_frames": 97,
        "decode_ok": true,
        "duration_s": 2.425,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "c66b9e1909d840d1dce8bfa279982dd86bff29d84f18521dbb43e4b0397cb16d",
        "source_sha256": {
          "third_person": "df2c41c8df278f28669a64f291756e9ed115b0b266dfb18f0204b1a096756a1b",
          "left_wrist": "e4f99156390e9417fff2a5391da85132c215e9c94453c9006a8a2072414c77a1",
          "right_wrist": "3e91e85867099607f1ecc49dc97e377e612eb7cd220b232ff84e76346ca880c6"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "f05b6b57e386",
          "task": "robotwin/place-shoe-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 97,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 97,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 98,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 97,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-43",
      "variant": "kinex",
      "task_key": "task03/43",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "43",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 3.6,
        "step_limit": 400,
        "steps": 90,
        "success": true,
        "valid": true
      },
      "steps": 90,
      "simulation_time_s": 3.6,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 3.6,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Use the left arm to press the shiny red plastic stapler.",
      "native_instruction": "Use the left arm to press the shiny red plastic stapler.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9607127290757794,
        "cache_read_input_tokens": 480512,
        "cache_reported_input_tokens": 500162,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 500162,
        "cached_input_tokens": 480512,
        "cost_usd": null,
        "fresh_input_tokens": 19650,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 500162,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 480512,
        "known_fresh_input_tokens": 19650,
        "known_input_tokens": 500162,
        "known_output_tokens": 3439,
        "known_reasoning_output_tokens": 320,
        "output_tokens": 3439,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 320,
        "reasoning_reported_output_tokens": 3439,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 23,
          "cache_write_input_tokens": 23,
          "cache_write_reported_input_tokens": 23,
          "cached_input_tokens": 23,
          "fresh_input_tokens": 23,
          "input_tokens": 23,
          "output_tokens": 23,
          "reasoning_output_tokens": 23,
          "reasoning_reported_output_tokens": 23
        },
        "request_attempts": 23,
        "response_count": 23,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 503601,
        "uncached_input_tokens": 19650,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11634,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11674,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11807,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 15286,
            "output_tokens": 43,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15232,
            "input_tokens": 15382,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15360,
            "input_tokens": 18404,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18304,
            "input_tokens": 19651,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 19815,
            "output_tokens": 94,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 20840,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 21092,
            "output_tokens": 365,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21853,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22124,
            "output_tokens": 357,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23261,
            "output_tokens": 124,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 24544,
            "output_tokens": 227,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 24968,
            "output_tokens": 901,
            "reasoning_output_tokens": 117
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25890,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26598,
            "output_tokens": 128,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 26753,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27247,
            "output_tokens": 108,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27489,
            "output_tokens": 50,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27561,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27936,
            "output_tokens": 393,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28353,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 29
      },
      "analysis": {
        "observed": "Native success after 90 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-43/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-43/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-43/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-43/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-43/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-43/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-43/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-43/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-43/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-43/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-43/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-43/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-43/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-43/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-43/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-43/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-43/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-43/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_press.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-43/workspace/.kinex/tools/cartesian_press.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/press-stapler.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-43/workspace/.kinex/memos/press-stapler.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 54,
        "observed_images": 3
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 37,
        "published_frames": 37,
        "decode_ok": true,
        "duration_s": 0.925,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "61b14a423b900e5208129a88743e73aba9af1ef47a6e22eb2953218f84d3ea06",
        "source_sha256": {
          "third_person": "72f0e8e8720671679292783c24e120fa5f0b07f9d5ec97a18b4345b11338303a",
          "left_wrist": "be482ec207842ccd2a3e639ad93e2b515ca860531f05621c722c9ddba35aebd7",
          "right_wrist": "246e7447c9ec35e315dd7efe5b0d3e71127bb32ff87050989c316f1adb23c747"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ef2badc19939",
          "task": "robotwin/press-stapler-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 38,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 37,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 38,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 37,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 38,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 37,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 3600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-44",
      "variant": "kinex",
      "task_key": "task03/44",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "44",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 13.6,
        "step_limit": 400,
        "steps": 340,
        "success": true,
        "valid": true
      },
      "steps": 340,
      "simulation_time_s": 13.6,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 13.6,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Catch the tabletop paymentsign, pick it up and rotate it, then set it back down.",
      "native_instruction": "Catch the tabletop paymentsign, pick it up and rotate it, then set it back down.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9630727801707696,
        "cache_read_input_tokens": 934144,
        "cache_reported_input_tokens": 969962,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 969962,
        "cached_input_tokens": 934144,
        "cost_usd": null,
        "fresh_input_tokens": 35818,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 969962,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 934144,
        "known_fresh_input_tokens": 35818,
        "known_input_tokens": 969962,
        "known_output_tokens": 7532,
        "known_reasoning_output_tokens": 1918,
        "output_tokens": 7532,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1918,
        "reasoning_reported_output_tokens": 7532,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 977494,
        "uncached_input_tokens": 35818,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11635,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11675,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11808,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 15757,
            "output_tokens": 263,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 19699,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20946,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20864,
            "input_tokens": 21176,
            "output_tokens": 164,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 23584,
            "output_tokens": 199,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 25380,
            "output_tokens": 383,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 26497,
            "output_tokens": 618,
            "reasoning_output_tokens": 292
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27379,
            "output_tokens": 625,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28342,
            "output_tokens": 1044,
            "reasoning_output_tokens": 210
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 29409,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30152,
            "output_tokens": 71,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30252,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30716,
            "output_tokens": 115,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31196,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31606,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31616,
            "input_tokens": 32071,
            "output_tokens": 694,
            "reasoning_output_tokens": 364
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33695,
            "output_tokens": 153,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34208,
            "output_tokens": 49,
            "reasoning_output_tokens": 12
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34176,
            "input_tokens": 34279,
            "output_tokens": 438,
            "reasoning_output_tokens": 212
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35032,
            "output_tokens": 215,
            "reasoning_output_tokens": 109
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35611,
            "output_tokens": 86,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35719,
            "output_tokens": 417,
            "reasoning_output_tokens": 194
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36165,
            "output_tokens": 130,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36718,
            "output_tokens": 201,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37342,
            "output_tokens": 60,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37536,
            "output_tokens": 80,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38014,
            "output_tokens": 385,
            "reasoning_output_tokens": 105
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38428,
            "output_tokens": 485,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 38938,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 38997,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 46
      },
      "analysis": {
        "observed": "Native success after 340 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-44/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-44/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-44/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-44/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-44/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-44/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-44/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-44/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-44/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-44/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-44/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-44/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-44/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-44/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-44/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-44/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-44/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-44/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-44/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/rotate-qrcode.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-44/workspace/.kinex/memos/rotate-qrcode.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 81,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 137,
        "published_frames": 137,
        "decode_ok": true,
        "duration_s": 3.425,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "39ac76b9f382334b819f46b904b45290c1d6c68b466bf69f2e036bd3e9554e57",
        "source_sha256": {
          "third_person": "1bfa26178952c6a9b973f252b24635c3826322a998521a7b5ed6cdc09dbf14c4",
          "left_wrist": "6ff85c64a58813562d48c3f7c7fd67730e5dc5ad6a0fcd76b74e62c5ccbb0343",
          "right_wrist": "e443a7832bde5421a02806ac7e73150f69790c39001b868683c8c94ce83b881b"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "da614e73e544",
          "task": "robotwin/rotate-qrcode-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 137,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13600000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 137,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13600000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 138,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 137,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13600000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-45",
      "variant": "kinex",
      "task_key": "task03/45",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "45",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 9.32,
        "step_limit": 700,
        "steps": 233,
        "success": true,
        "valid": true
      },
      "steps": 233,
      "simulation_time_s": 9.32,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 9.32,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up the bottle with white cap with the left arm and shake.",
      "native_instruction": "Pick up the bottle with white cap with the left arm and shake.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9650292635122929,
        "cache_read_input_tokens": 603648,
        "cache_reported_input_tokens": 625523,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 625523,
        "cached_input_tokens": 603648,
        "cost_usd": null,
        "fresh_input_tokens": 21875,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 625523,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 603648,
        "known_fresh_input_tokens": 21875,
        "known_input_tokens": 625523,
        "known_output_tokens": 4293,
        "known_reasoning_output_tokens": 670,
        "output_tokens": 4293,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 670,
        "reasoning_reported_output_tokens": 4293,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 26,
          "cache_write_input_tokens": 26,
          "cache_write_reported_input_tokens": 26,
          "cached_input_tokens": 26,
          "fresh_input_tokens": 26,
          "input_tokens": 26,
          "output_tokens": 26,
          "reasoning_output_tokens": 26,
          "reasoning_reported_output_tokens": 26
        },
        "request_attempts": 26,
        "response_count": 26,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 629816,
        "uncached_input_tokens": 21875,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11634,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11674,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11807,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 15723,
            "output_tokens": 199,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15744,
            "input_tokens": 18804,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 20051,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20215,
            "output_tokens": 141,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 22219,
            "output_tokens": 70,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22499,
            "output_tokens": 320,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23214,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 23485,
            "output_tokens": 321,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24959,
            "output_tokens": 782,
            "reasoning_output_tokens": 169
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25600,
            "input_tokens": 25762,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 25899,
            "output_tokens": 101,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25856,
            "input_tokens": 26564,
            "output_tokens": 82,
            "reasoning_output_tokens": 44
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26496,
            "input_tokens": 26668,
            "output_tokens": 323,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26880,
            "input_tokens": 27312,
            "output_tokens": 402,
            "reasoning_output_tokens": 223
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27873,
            "output_tokens": 199,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28098,
            "output_tokens": 123,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28790,
            "output_tokens": 96,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28672,
            "input_tokens": 29459,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30028,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30260,
            "output_tokens": 76,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30358,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30805,
            "output_tokens": 535,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31363,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 34
      },
      "analysis": {
        "observed": "Native success after 233 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-45/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-45/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-45/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-45/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-45/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-45/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-45/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-45/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-45/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-45/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-45/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-45/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-45/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-45/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-45/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-45/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-45/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-45/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_left.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-45/workspace/.kinex/tools/cartesian_left.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/shake-bottle.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-45/workspace/.kinex/memos/shake-bottle.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 62,
        "observed_images": 6
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 94,
        "published_frames": 94,
        "decode_ok": true,
        "duration_s": 2.35,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "eec6814a56750d0e1fc11c2ac28608c71878e17f9a1de96631821a57eba0fec0",
        "source_sha256": {
          "third_person": "8105c5eeb0bcdaf8e36c3acf00bbe15e71521d1a11d9c6391bc9858178f7b9eb",
          "left_wrist": "5815e4f923961c99d8f1e97fa1112496fbe51540210aa1e4fcc389fd96ffbc08",
          "right_wrist": "49c0de595d63ed3adb2716c3bb0de74e9340368f038a397ca9cbf2d95a64abb9"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "d7d4468e7d0c",
          "task": "robotwin/shake-bottle-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 95,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 94,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9320000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 95,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 94,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9320000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 95,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 94,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 9320000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-46",
      "variant": "kinex",
      "task_key": "task03/46",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "46",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 5.5200000000000005,
        "step_limit": 700,
        "steps": 138,
        "success": true,
        "valid": true
      },
      "steps": 138,
      "simulation_time_s": 5.5200000000000005,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 5.5200000000000005,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the bottle with white cap and shake it horizontally.",
      "native_instruction": "Grab the bottle with white cap and shake it horizontally.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9553317567028937,
        "cache_read_input_tokens": 1150464,
        "cache_reported_input_tokens": 1204256,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1204256,
        "cached_input_tokens": 1150464,
        "cost_usd": null,
        "fresh_input_tokens": 53792,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1204256,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1150464,
        "known_fresh_input_tokens": 53792,
        "known_input_tokens": 1204256,
        "known_output_tokens": 5355,
        "known_reasoning_output_tokens": 621,
        "output_tokens": 5355,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 621,
        "reasoning_reported_output_tokens": 5355,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 32,
          "cache_write_input_tokens": 32,
          "cache_write_reported_input_tokens": 32,
          "cached_input_tokens": 32,
          "fresh_input_tokens": 32,
          "input_tokens": 32,
          "output_tokens": 32,
          "reasoning_output_tokens": 32,
          "reasoning_reported_output_tokens": 32
        },
        "request_attempts": 32,
        "response_count": 32,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1209611,
        "uncached_input_tokens": 53792,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11634,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11669,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11800,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11848,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12098,
            "output_tokens": 192,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 15933,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15872,
            "input_tokens": 17194,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17152,
            "input_tokens": 17390,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17280,
            "input_tokens": 17582,
            "output_tokens": 71,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17536,
            "input_tokens": 18193,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 18445,
            "output_tokens": 153,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18432,
            "input_tokens": 40578,
            "output_tokens": 353,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 42151,
            "output_tokens": 151,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 44250,
            "output_tokens": 173,
            "reasoning_output_tokens": 147
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44441,
            "output_tokens": 290,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 45274,
            "output_tokens": 60,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 46513,
            "output_tokens": 302,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 47424,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47360,
            "input_tokens": 48237,
            "output_tokens": 926,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49024,
            "input_tokens": 49184,
            "output_tokens": 113,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49152,
            "input_tokens": 49795,
            "output_tokens": 100,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49792,
            "input_tokens": 50612,
            "output_tokens": 305,
            "reasoning_output_tokens": 65
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50816,
            "input_tokens": 51231,
            "output_tokens": 362,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51456,
            "input_tokens": 51861,
            "output_tokens": 347,
            "reasoning_output_tokens": 99
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52096,
            "input_tokens": 52235,
            "output_tokens": 86,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52224,
            "input_tokens": 52689,
            "output_tokens": 93,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52608,
            "input_tokens": 53157,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52992,
            "input_tokens": 53579,
            "output_tokens": 118,
            "reasoning_output_tokens": 23
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53632,
            "input_tokens": 53836,
            "output_tokens": 70,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53888,
            "input_tokens": 53968,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53888,
            "input_tokens": 54414,
            "output_tokens": 602,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54912,
            "input_tokens": 55041,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 38
      },
      "analysis": {
        "observed": "Native success after 138 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-46/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-46/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-46/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-46/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-46/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-46/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-46/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-46/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-46/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-46/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-46/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-46/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-46/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-46/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-46/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-46/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-46/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-46/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-46/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/robotwin-arx5.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-46/workspace/.kinex/memos/robotwin-arx5.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 72,
        "observed_images": 5
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 56,
        "published_frames": 56,
        "decode_ok": true,
        "duration_s": 1.4,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "01c4113d9f8f19a4ff7cb57684e73f4bdf06d2ebb2d1272942c5b0b923918d3b",
        "source_sha256": {
          "third_person": "0258226238237fe2d0aa038b16638112264fc3a531e6c5570a0f8e0dbd15a23b",
          "left_wrist": "566642086d30403fcdf0320ba1bb052243bc97ce4101f0b8430e4ada4cd29150",
          "right_wrist": "2df3cba861923373feaff67475fcb956373f3d183e41ddd9dded394c23cf2665"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "503578721507",
          "task": "robotwin/shake-bottle-horizontally-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 57,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 56,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 5520000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 57,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 56,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 5520000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 57,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 56,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 5520000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-47",
      "variant": "kinex",
      "task_key": "task03/47",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "47",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 17.8,
        "step_limit": 800,
        "steps": 445,
        "success": true,
        "valid": true
      },
      "steps": 445,
      "simulation_time_s": 17.8,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 17.8,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Bring red block and green block to the center, then position green block above red block.",
      "native_instruction": "Bring red block and green block to the center, then position green block above red block.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9509136389007391,
        "cache_read_input_tokens": 1011968,
        "cache_reported_input_tokens": 1064206,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1064206,
        "cached_input_tokens": 1011968,
        "cost_usd": null,
        "fresh_input_tokens": 52238,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1064206,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1011968,
        "known_fresh_input_tokens": 52238,
        "known_input_tokens": 1064206,
        "known_output_tokens": 6655,
        "known_reasoning_output_tokens": 1500,
        "output_tokens": 6655,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1500,
        "reasoning_reported_output_tokens": 6655,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 35,
          "cache_write_input_tokens": 35,
          "cache_write_reported_input_tokens": 35,
          "cached_input_tokens": 35,
          "fresh_input_tokens": 35,
          "input_tokens": 35,
          "output_tokens": 35,
          "reasoning_output_tokens": 35,
          "reasoning_reported_output_tokens": 35
        },
        "request_attempts": 35,
        "response_count": 35,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1070861,
        "uncached_input_tokens": 52238,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11639,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11679,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11812,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 15270,
            "output_tokens": 246,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15360,
            "input_tokens": 19192,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 20761,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20736,
            "input_tokens": 20991,
            "output_tokens": 166,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20992,
            "input_tokens": 23587,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 23839,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23680,
            "input_tokens": 24110,
            "output_tokens": 392,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 25008,
            "output_tokens": 261,
            "reasoning_output_tokens": 136
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 27605,
            "output_tokens": 364,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27776,
            "input_tokens": 28393,
            "output_tokens": 1525,
            "reasoning_output_tokens": 290
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29939,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30608,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30732,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 31360,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31992,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32526,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33163,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33933,
            "output_tokens": 457,
            "reasoning_output_tokens": 340
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 35033,
            "output_tokens": 176,
            "reasoning_output_tokens": 119
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35787,
            "output_tokens": 66,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36355,
            "output_tokens": 162,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 37158,
            "output_tokens": 79,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37120,
            "input_tokens": 37814,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37632,
            "input_tokens": 38352,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38984,
            "output_tokens": 445,
            "reasoning_output_tokens": 168
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 39674,
            "output_tokens": 230,
            "reasoning_output_tokens": 129
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39808,
            "input_tokens": 40546,
            "output_tokens": 57,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40737,
            "output_tokens": 69,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 40868,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41061,
            "output_tokens": 437,
            "reasoning_output_tokens": 130
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41525,
            "output_tokens": 625,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41984,
            "input_tokens": 42173,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 47
      },
      "analysis": {
        "observed": "Native success after 445 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-47/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-47/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-47/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-47/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-47/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-47/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-47/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-47/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-47/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-47/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-47/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-47/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-47/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-47/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-47/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-47/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-47/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-47/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/stack_blocks.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-47/workspace/.kinex/tools/stack_blocks.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/stack-blocks.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-47/workspace/.kinex/memos/stack-blocks.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 84,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 179,
        "published_frames": 179,
        "decode_ok": true,
        "duration_s": 4.475,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "5c228bdc4452837eb05f22ceae620e0d31ba08dfd2fc1c7336c119e733666385",
        "source_sha256": {
          "third_person": "e2c5a5e8740aa5a81e693a1ba0a27949edd861ab6f95779831dfdb1e3805644c",
          "left_wrist": "91daa250210ebe418758322407f89e2a201885300d4d7b706cfe4fcfc8ad1a09",
          "right_wrist": "9a692f58aae4dcd9c2a1f7faa58a0a048d2e150d6244c176159a05c23e42f0d5"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "2ccc9b77b19e",
          "task": "robotwin/stack-blocks-two-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 180,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 179,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17800000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 180,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 179,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17800000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 180,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 179,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 17800000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-48",
      "variant": "kinex",
      "task_key": "task03/48",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "48",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 1
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 21.16,
        "step_limit": 900,
        "steps": 529,
        "success": true,
        "valid": true
      },
      "steps": 529,
      "simulation_time_s": 21.16,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 21.16,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Stack the small smooth brown-rimmed bowl directly over the smooth bowl with glossy finish.",
      "native_instruction": "Stack the small smooth brown-rimmed bowl directly over the smooth bowl with glossy finish.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9759635552053721,
        "cache_read_input_tokens": 1147008,
        "cache_reported_input_tokens": 1175257,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1175257,
        "cached_input_tokens": 1147008,
        "cost_usd": null,
        "fresh_input_tokens": 28249,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1175257,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1147008,
        "known_fresh_input_tokens": 28249,
        "known_input_tokens": 1175257,
        "known_output_tokens": 7103,
        "known_reasoning_output_tokens": 1519,
        "output_tokens": 7103,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1519,
        "reasoning_reported_output_tokens": 7103,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 43,
          "cache_write_input_tokens": 43,
          "cache_write_reported_input_tokens": 43,
          "cached_input_tokens": 43,
          "fresh_input_tokens": 43,
          "input_tokens": 43,
          "output_tokens": 43,
          "reasoning_output_tokens": 43,
          "reasoning_reported_output_tokens": 43
        },
        "request_attempts": 43,
        "response_count": 43,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1182360,
        "uncached_input_tokens": 28249,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11642,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11682,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11815,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11865,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 14070,
            "output_tokens": 193,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14080,
            "input_tokens": 17903,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 19152,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19072,
            "input_tokens": 19348,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 19538,
            "output_tokens": 119,
            "reasoning_output_tokens": 78
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 20197,
            "output_tokens": 151,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 22560,
            "output_tokens": 111,
            "reasoning_output_tokens": 72
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22734,
            "output_tokens": 59,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23482,
            "output_tokens": 161,
            "reasoning_output_tokens": 13
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24553,
            "output_tokens": 149,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 24815,
            "output_tokens": 214,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25298,
            "output_tokens": 318,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25472,
            "input_tokens": 25926,
            "output_tokens": 311,
            "reasoning_output_tokens": 143
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26427,
            "output_tokens": 161,
            "reasoning_output_tokens": 122
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 27283,
            "output_tokens": 404,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 27839,
            "output_tokens": 268,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28488,
            "output_tokens": 281,
            "reasoning_output_tokens": 191
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28891,
            "output_tokens": 98,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28800,
            "input_tokens": 29783,
            "output_tokens": 81,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 29882,
            "output_tokens": 925,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30592,
            "input_tokens": 30830,
            "output_tokens": 90,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31226,
            "output_tokens": 97,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31516,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31835,
            "output_tokens": 146,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32287,
            "output_tokens": 51,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32534,
            "output_tokens": 77,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32920,
            "output_tokens": 397,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33456,
            "output_tokens": 168,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33931,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34182,
            "output_tokens": 75,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34508,
            "output_tokens": 140,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34897,
            "output_tokens": 317,
            "reasoning_output_tokens": 38
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35072,
            "input_tokens": 35554,
            "output_tokens": 291,
            "reasoning_output_tokens": 214
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36094,
            "output_tokens": 70,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 36298,
            "output_tokens": 75,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36435,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 36743,
            "output_tokens": 628,
            "reasoning_output_tokens": 59
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37397,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37441,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 50
      },
      "analysis": {
        "observed": "Native success after 529 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-48/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-48/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-48/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-48/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-48/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-48/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-48/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-48/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-48/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-48/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-48/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-48/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-48/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-48/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-48/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-48/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-48/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-48/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/bowl_motion.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-48/workspace/.kinex/tools/bowl_motion.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/stack_bowls.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-48/workspace/.kinex/memos/stack_bowls.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 95,
        "observed_images": 15
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 213,
        "published_frames": 213,
        "decode_ok": true,
        "duration_s": 5.325,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "f246ce2b26384994b5850e88abda32ce46df7ae26fe5b4d63e102e12be05ff45",
        "source_sha256": {
          "third_person": "1d46db5151e5442f32e7002a6fbe1e3d85a6e75826cced5db59e6fff39f450c2",
          "left_wrist": "0bdf13cfd9fae0e4af5dc8d5067e870e7371569302401880f444dd6b6cbb48b7",
          "right_wrist": "95ac47731321d4a8f47d19c2d256b6bd66432f860e75fdf1779aa3b81fa84fca"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "86644f4e0105",
          "task": "robotwin/stack-bowls-two-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 213,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 213,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 21160000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 213,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 213,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 21160000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 213,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 213,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 21160000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-49",
      "variant": "kinex",
      "task_key": "task03/49",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "49",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 12.16,
        "step_limit": 400,
        "steps": 304,
        "success": true,
        "valid": true
      },
      "steps": 304,
      "simulation_time_s": 12.16,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 12.16,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Stamp Green using the block-shaped seal with angled edges",
      "native_instruction": "Stamp Green using the block-shaped seal with angled edges",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9352915385539107,
        "cache_read_input_tokens": 669824,
        "cache_reported_input_tokens": 716166,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 716166,
        "cached_input_tokens": 669824,
        "cost_usd": null,
        "fresh_input_tokens": 46342,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 716166,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 669824,
        "known_fresh_input_tokens": 46342,
        "known_input_tokens": 716166,
        "known_output_tokens": 5569,
        "known_reasoning_output_tokens": 1073,
        "output_tokens": 5569,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1073,
        "reasoning_reported_output_tokens": 5569,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 721735,
        "uncached_input_tokens": 46342,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 11632,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11680,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11811,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11869,
            "output_tokens": 21,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11904,
            "output_tokens": 22,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11940,
            "output_tokens": 192,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 15775,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15616,
            "input_tokens": 17022,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16896,
            "input_tokens": 17601,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 17765,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17664,
            "input_tokens": 17957,
            "output_tokens": 75,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18242,
            "output_tokens": 134,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18048,
            "input_tokens": 19077,
            "output_tokens": 283,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 20074,
            "output_tokens": 184,
            "reasoning_output_tokens": 145
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20724,
            "output_tokens": 500,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20608,
            "input_tokens": 21767,
            "output_tokens": 419,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 22414,
            "output_tokens": 62,
            "reasoning_output_tokens": 24
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 23171,
            "output_tokens": 950,
            "reasoning_output_tokens": 182
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23040,
            "input_tokens": 24142,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23936,
            "input_tokens": 24564,
            "output_tokens": 110,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24448,
            "input_tokens": 25044,
            "output_tokens": 197,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24832,
            "input_tokens": 25577,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26015,
            "output_tokens": 478,
            "reasoning_output_tokens": 137
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25984,
            "input_tokens": 26846,
            "output_tokens": 119,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26752,
            "input_tokens": 27343,
            "output_tokens": 67,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27783,
            "output_tokens": 168,
            "reasoning_output_tokens": 101
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27648,
            "input_tokens": 28321,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28761,
            "output_tokens": 340,
            "reasoning_output_tokens": 240
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29478,
            "output_tokens": 127,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29739,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 29795,
            "output_tokens": 46,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 8960,
            "input_tokens": 29865,
            "output_tokens": 580,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30468,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 37
      },
      "analysis": {
        "observed": "Native success after 304 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-49/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-49/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-49/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-49/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-49/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-49/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-49/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-49/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-49/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-49/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-49/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-49/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-49/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-49/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-49/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-49/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-49/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-49/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-49/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/stamp-seal.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-49/workspace/.kinex/memos/stamp-seal.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 72,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 123,
        "published_frames": 123,
        "decode_ok": true,
        "duration_s": 3.075,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "7b50ea77b141ebf3b68ee84db4805d5b19f150fcb0a12c81ab0635fead04cf42",
        "source_sha256": {
          "third_person": "f70b44308abb9bb34f072086a32994be96b75004d90ef6e85241b0fabb913b2a",
          "left_wrist": "71eeaec761b5b399f2bc1459100af360710079b71ce2d8d4e86f0bc0a525f376",
          "right_wrist": "533ca33f900104b9cd5d52ce91459e69da4bc408a87d87b83b011fab1738f03c"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "ce09111d8d7b",
          "task": "robotwin/stamp-seal-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 123,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 123,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12160000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 123,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 123,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12160000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 123,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 123,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 12160000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-kinex-50",
      "variant": "kinex",
      "task_key": "task03/50",
      "family": "task03",
      "source_family": "task03",
      "provider": "rosbridge",
      "slot": "50",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 13.8,
        "step_limit": 400,
        "steps": 345,
        "success": false,
        "valid": true
      },
      "steps": 345,
      "simulation_time_s": 13.8,
      "wall_time_s": null,
      "timing": {
        "simulation_time_s": 13.8,
        "wall_time_s": null,
        "wall_time_source": "not aggregated; see raw session timestamps"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Click the flat gray switch with the right arm",
      "native_instruction": "Click the flat gray switch with the right arm",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9545162810961171,
        "cache_read_input_tokens": 694528,
        "cache_reported_input_tokens": 727623,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 727623,
        "cached_input_tokens": 694528,
        "cost_usd": null,
        "fresh_input_tokens": 33095,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 727623,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 694528,
        "known_fresh_input_tokens": 33095,
        "known_input_tokens": 727623,
        "known_output_tokens": 5722,
        "known_reasoning_output_tokens": 1274,
        "output_tokens": 5722,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1274,
        "reasoning_reported_output_tokens": 5722,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 29,
          "cache_write_input_tokens": 29,
          "cache_write_reported_input_tokens": 29,
          "cached_input_tokens": 29,
          "fresh_input_tokens": 29,
          "input_tokens": 29,
          "output_tokens": 29,
          "reasoning_output_tokens": 29,
          "reasoning_reported_output_tokens": 29
        },
        "request_attempts": 29,
        "response_count": 29,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 733345,
        "uncached_input_tokens": 33095,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11627,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11662,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11520,
            "input_tokens": 11795,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 14744,
            "output_tokens": 197,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14848,
            "input_tokens": 17830,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 19077,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 19241,
            "output_tokens": 50,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 19702,
            "output_tokens": 59,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 20301,
            "output_tokens": 125,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 21559,
            "output_tokens": 224,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 22587,
            "output_tokens": 136,
            "reasoning_output_tokens": 20
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 24232,
            "output_tokens": 499,
            "reasoning_output_tokens": 134
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24576,
            "input_tokens": 25168,
            "output_tokens": 232,
            "reasoning_output_tokens": 193
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25216,
            "input_tokens": 25866,
            "output_tokens": 442,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26469,
            "output_tokens": 319,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26624,
            "input_tokens": 27128,
            "output_tokens": 482,
            "reasoning_output_tokens": 140
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27520,
            "input_tokens": 28189,
            "output_tokens": 927,
            "reasoning_output_tokens": 165
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29139,
            "output_tokens": 89,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29434,
            "output_tokens": 84,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29895,
            "output_tokens": 126,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29824,
            "input_tokens": 30400,
            "output_tokens": 158,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 30934,
            "output_tokens": 103,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30848,
            "input_tokens": 31413,
            "output_tokens": 203,
            "reasoning_output_tokens": 132
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 31991,
            "output_tokens": 154,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32517,
            "output_tokens": 105,
            "reasoning_output_tokens": 35
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32997,
            "output_tokens": 228,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33671,
            "output_tokens": 77,
            "reasoning_output_tokens": 39
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33536,
            "input_tokens": 33770,
            "output_tokens": 491,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34285,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "model_tool_calls": 36
      },
      "analysis": {
        "observed": "Native failure after 345 rosbridge control frames. One valid episode with the native unseen instruction; no valid-outcome retry.",
        "execution": "Previous native-instruction campaign; replaced in the current panel by RLE revised instructions. Original evidence remains unchanged."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-kinex-50/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-kinex-50/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-kinex-50/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-kinex-50/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-kinex-50/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-kinex-50/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-kinex-50/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-kinex-50/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-kinex-50/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-kinex-50/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-kinex-50/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-kinex-50/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-kinex-50/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-kinex-50/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-kinex-50/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-kinex-50/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_session": "experiments/robotwin-hard-20261101/hard-kinex-50/session.jsonl",
        "trace": "experiments/robotwin-hard-20261101/hard-kinex-50/trace.json"
      },
      "resources": [
        {
          "name": ".kinex/tools/right_cartesian.py",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-50/workspace/.kinex/tools/right_cartesian.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/memos/turn_switch.md",
          "file": "experiments/robotwin-hard-20261101/hard-kinex-50/workspace/.kinex/memos/turn_switch.md",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 67,
        "observed_images": 10
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "First fully initializable native Hard seed, ascending from 0, with native unseen instruction available. Do not filter expert success. Freeze selected seed before any model query."
      },
      "vla_calls": 0,
      "media": {
        "speed": 4,
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "native_frames": 139,
        "published_frames": 139,
        "decode_ok": true,
        "duration_s": 3.475,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "3279c773d55b54068f469b088961776a52d683fd7469a90d622fe41963d73efc",
        "source_sha256": {
          "third_person": "fccd6f85c7695c4a9fffa820a2ae73b5842497965617ad6e34363d9769672c27",
          "left_wrist": "dc609f136db8af5f4cc033f47606897e9b9b49ca503914cb0943226f8f71c58e",
          "right_wrist": "066caeceb9fb4653947d6dff10775eae6d91d8ccd00516e30ab59066346f6a9b"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "046cde8d7ea6",
          "task": "robotwin/turn-switch-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 140,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 139,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13800000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 140,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 139,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13800000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 140,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 139,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 13800000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "dataset_revision": "c833b470484d05a4afda77fbb6bb8cf24057d4d9"
    },
    {
      "id": "hard-clean-01",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/01",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "01",
      "seed": 100001,
      "source_episode_id": "01-handover_block",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 800
      },
      "steps": 800,
      "simulation_time_s": 181.53600862249732,
      "wall_time_s": 83.21375876,
      "timing": {
        "wall_time_s": 83.21375876,
        "simulation_time_s": 181.53600862249732,
        "physics_steps": 45384,
        "wall_time_start_ns": 1793551295473118594,
        "wall_time_end_ns": 1793551378686880088,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
      "native_instruction": "Grab the red block using the left arm, then transfer it to the right arm and place it on the blue pad.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 800 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-01/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-01/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-01/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-01/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-01/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-01/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-01/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-01/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-01/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-01/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-01/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-01/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-01/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-01/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-01/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-01/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-01/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-01/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 25,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 20.0,
        "native_frames": 800,
        "published_frames": 800,
        "decode_ok": true,
        "sha256": "a80fa3e5878a5e13962a9d62c132963b69050eb8cbcf351465eabd6ba93fdb9c",
        "source_sha256": "5bcf82a3c45732b560ee69d55e55bdacbff1a0bb96f294b2b756300e4c54da0f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-02",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/02",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "02",
      "seed": 100000,
      "source_episode_id": "02-place_dual_shoes",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 600
      },
      "steps": 600,
      "simulation_time_s": 111.51200529653579,
      "wall_time_s": 66.067087928,
      "timing": {
        "wall_time_s": 66.067087928,
        "simulation_time_s": 111.51200529653579,
        "physics_steps": 27878,
        "wall_time_start_ns": 1793551294712408702,
        "wall_time_end_ns": 1793551360779499450,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Pick up two the walking shoe with grip sole, tips left, into the rectangular Nike box for shoes.",
      "native_instruction": "Pick up two the walking shoe with grip sole, tips left, into the rectangular Nike box for shoes.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 600 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-02/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-02/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-02/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-02/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-02/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-02/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-02/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-02/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-02/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-02/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-02/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-02/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-02/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-02/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-02/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-02/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-02/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-02/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 15.0,
        "native_frames": 600,
        "published_frames": 600,
        "decode_ok": true,
        "sha256": "81b85875cabda0951c3e044baa92a54eb17ff9b107063606d8f880588148b3be",
        "source_sha256": "f5d9896dad38977dd468e36136cdb48eab60225ab6b2fd70ffd7b8dc2c414717",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-03",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/03",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "03",
      "seed": 100000,
      "source_episode_id": "03-put_bottles_dustbin",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1700
      },
      "steps": 1700,
      "simulation_time_s": 488.1760231871158,
      "wall_time_s": 172.159274475,
      "timing": {
        "wall_time_s": 172.159274475,
        "simulation_time_s": 488.1760231871158,
        "physics_steps": 122044,
        "wall_time_start_ns": 1793551403584840789,
        "wall_time_end_ns": 1793551575744118485,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
      "native_instruction": "Use the arms to move the yellow bottle with elongated cylinder body, the bottle with blue highlights, and the bottle with red screw cap inside the simple black garbage bin.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 1700 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-03/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-03/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-03/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-03/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-03/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-03/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-03/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-03/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-03/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-03/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-03/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-03/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-03/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-03/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-03/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-03/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-03/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-03/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1700,
        "max_control_steps": 1700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 54,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 42.5,
        "native_frames": 1700,
        "published_frames": 1700,
        "decode_ok": true,
        "sha256": "0a61ada3e245e163d43efee54decfc78b0fc82aa08b9c861723b5763b789c492",
        "source_sha256": "4c02873a49d7188ed7da91e3087c7c07dc6a2e787ca9f0626d256c9a048fa615",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184,
          1216,
          1248,
          1280,
          1312,
          1344,
          1376,
          1408,
          1440,
          1472,
          1504,
          1536,
          1568,
          1600,
          1632,
          1664,
          1696
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-04",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/04",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "04",
      "seed": 100004,
      "source_episode_id": "04-scan_object",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 129.02800612850115,
      "wall_time_s": 61.127255626,
      "timing": {
        "wall_time_s": 61.127255626,
        "simulation_time_s": 129.02800612850115,
        "physics_steps": 32257,
        "wall_time_start_ns": 1793551423940541041,
        "wall_time_end_ns": 1793551485067799003,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use one arm for the box with green leafy patterns and the other for the black scanner with ergonomic handle, then scan.",
      "native_instruction": "Use one arm for the box with green leafy patterns and the other for the black scanner with ergonomic handle, then scan.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 500 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-04/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-04/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-04/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-04/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-04/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-04/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-04/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-04/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-04/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-04/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-04/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-04/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-04/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-04/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-04/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-04/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-04/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-04/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100004,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "ff5740a43b11f33d67682b9e53a81e6e5a85f2f6d22f465f9fd9051530dec386",
        "source_sha256": "5023c3e66609eeba377b2b9e5e17e8e96b0774f65e4de20a087454e01679fb09",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-05",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/05",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "05",
      "seed": 100000,
      "source_episode_id": "05-stack_bowls_three",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 604
      },
      "steps": 604,
      "simulation_time_s": 174.34800828108564,
      "wall_time_s": 72.926394029,
      "timing": {
        "wall_time_s": 72.926394029,
        "simulation_time_s": 174.34800828108564,
        "physics_steps": 43587,
        "wall_time_start_ns": 1793551518462089159,
        "wall_time_end_ns": 1793551591388486155,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
      "native_instruction": "Stack the first the smooth bowl with glossy finish, then the second, then the third.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 604 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-05/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-05/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-05/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-05/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-05/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-05/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-05/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-05/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-05/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-05/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-05/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-05/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-05/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-05/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-05/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-05/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-05/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-05/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 19,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 15.125,
        "native_frames": 605,
        "published_frames": 605,
        "decode_ok": true,
        "sha256": "8cf359b7e25e728687f4ac46bb06f96e904e0dcc7950355fc2da41ce38010929",
        "source_sha256": "8565e110e4b0dcd03a7994644ef1c517f98fc4aa2d90133b3507ea65e20f8704",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-06",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/06",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "06",
      "seed": 100000,
      "source_episode_id": "06-stack_blocks_three",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 392.6760186511092,
      "wall_time_s": 132.283480687,
      "timing": {
        "wall_time_s": 132.283480687,
        "simulation_time_s": 392.6760186511092,
        "physics_steps": 98169,
        "wall_time_start_ns": 1793551609487098859,
        "wall_time_end_ns": 1793551741770582182,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
      "native_instruction": "Build a tower with red block at the bottom, green block in the middle, and blue block on top.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 1200 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-06/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-06/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-06/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-06/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-06/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-06/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-06/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-06/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-06/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-06/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-06/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-06/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-06/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-06/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-06/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-06/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-06/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-06/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "65521e9fb04d76723e5af37e77019847110c099f39b33ae63ae29828cb7e04c9",
        "source_sha256": "3aa23b3d68cba96dabb4d240a92e5d59877a0157380d0625018d2525d44fd767",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-07",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/07",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "07",
      "seed": 100000,
      "source_episode_id": "07-hanging_mug",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 900
      },
      "steps": 900,
      "simulation_time_s": 227.92801082599908,
      "wall_time_s": 100.234778203,
      "timing": {
        "wall_time_s": 100.234778203,
        "simulation_time_s": 227.92801082599908,
        "physics_steps": 56982,
        "wall_time_start_ns": 1793551627069070325,
        "wall_time_end_ns": 1793551727303851670,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
      "native_instruction": "Lift the beige mug from the table, spin it, set it down, and hang it on the rack with straight and slanted rods.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 900 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-07/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-07/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-07/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-07/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-07/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-07/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-07/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-07/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-07/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-07/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-07/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-07/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-07/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-07/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-07/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-07/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-07/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-07/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 29,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 22.5,
        "native_frames": 900,
        "published_frames": 900,
        "decode_ok": true,
        "sha256": "d2408e181ab99d14a639178c30467b28650c7f9991356c15c12620e1ab3c3ee4",
        "source_sha256": "a68ae5a7e0f77ad4835b215c1dbc211aee70053c28fe625ff8eb4c7f0dcf5c7e",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-08",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/08",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "08",
      "seed": 100009,
      "source_episode_id": "08-put_object_cabinet",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 148.4640070516616,
      "wall_time_s": 82.812127386,
      "timing": {
        "wall_time_s": 82.812127386,
        "simulation_time_s": 148.4640070516616,
        "physics_steps": 37116,
        "wall_time_start_ns": 1793551786736471854,
        "wall_time_end_ns": 1793551869548601257,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Open the slanted wooden cabinet's drawer and set the rectangular playingcards case with patterns inside with the other arm.",
      "native_instruction": "Open the slanted wooden cabinet's drawer and set the rectangular playingcards case with patterns inside with the other arm.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 700 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-08/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-08/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-08/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-08/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-08/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-08/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-08/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-08/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-08/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-08/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-08/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-08/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-08/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-08/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-08/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-08/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-08/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-08/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100009,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "cceb940094b59d9148c69939f05dffbc088937ae142b76f537198622673555ca",
        "source_sha256": "8c36e6fdb1ab10b789cd5ba6b35b87c9d5f804e097a8521b771d3a6740590ba1",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-09",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/09",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "09",
      "seed": 100000,
      "source_episode_id": "09-blocks_ranking_size",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 442
      },
      "steps": 442,
      "simulation_time_s": 125.66000596852973,
      "wall_time_s": 49.564632283,
      "timing": {
        "wall_time_s": 49.564632283,
        "simulation_time_s": 125.66000596852973,
        "physics_steps": 31415,
        "wall_time_start_ns": 1793551775877116494,
        "wall_time_end_ns": 1793551825441751722,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Place large block, medium block, and small block in order at the table's center.",
      "native_instruction": "Place large block, medium block, and small block in order at the table's center.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 442 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-09/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-09/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-09/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-09/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-09/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-09/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-09/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-09/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-09/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-09/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-09/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-09/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-09/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-09/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-09/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-09/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-09/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-09/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 14,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 11.075,
        "native_frames": 443,
        "published_frames": 443,
        "decode_ok": true,
        "sha256": "c793bd03f55821d7b14d2c4513ae4b1906e63fa016c3cc425e2afd77052c6e9a",
        "source_sha256": "ca6565d139c3ee42f65cbcb6fc1a2676c2393d39e50d884df1ba7a47e99bd0b5",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-10",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/10",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "10",
      "seed": 100000,
      "source_episode_id": "10-lift_pot",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 66.80800317320973,
      "wall_time_s": 54.632876051,
      "timing": {
        "wall_time_s": 54.632876051,
        "simulation_time_s": 66.80800317320973,
        "physics_steps": 16702,
        "wall_time_start_ns": 1793551859380192236,
        "wall_time_end_ns": 1793551914013070286,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Lift the dark gray cooking kitchenpot using both arms",
      "native_instruction": "Lift the dark gray cooking kitchenpot using both arms",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-10/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-10/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-10/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-10/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-10/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-10/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-10/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-10/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-10/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-10/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-10/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-10/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-10/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-10/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-10/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-10/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-10/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-10/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "c156a90598d90388e60e1630570e15989cf0a0a7d57479f011fcbcebcfae0a8b",
        "source_sha256": "a3fa83d3c4db15990363e58c76cb4ce1b71dc6f7338f87b9d96cc0aa36050c24",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-11",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/11",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "11",
      "seed": 100000,
      "source_episode_id": "11-adjust_bottle",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 95.0680045154877,
      "wall_time_s": 49.074998827,
      "timing": {
        "wall_time_s": 49.074998827,
        "simulation_time_s": 95.0680045154877,
        "physics_steps": 23767,
        "wall_time_start_ns": 1793551902295919747,
        "wall_time_end_ns": 1793551951370921597,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the brown bottle upright",
      "native_instruction": "Use the left arm to grab the brown bottle upright",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-11/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-11/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-11/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-11/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-11/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-11/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-11/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-11/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-11/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-11/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-11/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-11/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-11/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-11/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-11/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-11/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-11/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-11/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "c0cc30979309fdd53361c7faad33e3e9820be0cafa5e2bb25f1df9076390714f",
        "source_sha256": "d7e3e2ebc5c86da63db778027493b0f91fc32bf731169196cd809382e569ee86",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-12",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/12",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "12",
      "seed": 100000,
      "source_episode_id": "12-beat_block_hammer",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 103.7800049292855,
      "wall_time_s": 46.692232208,
      "timing": {
        "wall_time_s": 46.692232208,
        "simulation_time_s": 103.7800049292855,
        "physics_steps": 25945,
        "wall_time_start_ns": 1793551946233424039,
        "wall_time_end_ns": 1793551992925659505,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the hammer with black handle then beat block",
      "native_instruction": "Use the left arm to grab the hammer with black handle then beat block",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-12/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-12/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-12/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-12/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-12/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-12/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-12/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-12/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-12/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-12/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-12/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-12/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-12/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-12/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-12/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-12/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-12/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-12/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "9d2e0399ba967224c6b621de016393e0ebdccd669463a318708ba0a7e126ee9f",
        "source_sha256": "5bc82530f1dee4c0ec8e9f37be678540eb2c58927643555fa77ca2393403d301",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-13",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/13",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "13",
      "seed": 100000,
      "source_episode_id": "13-blocks_ranking_rgb",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1200
      },
      "steps": 1200,
      "simulation_time_s": 320.940015243832,
      "wall_time_s": 123.861012426,
      "timing": {
        "wall_time_s": 123.861012426,
        "simulation_time_s": 320.940015243832,
        "physics_steps": 80235,
        "wall_time_start_ns": 1793551983445789089,
        "wall_time_end_ns": 1793552107306804472,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
      "native_instruction": "Arrange red block, green block, and blue block side by side, starting with red block on the left.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 1200 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-13/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-13/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-13/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-13/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-13/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-13/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-13/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-13/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-13/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-13/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-13/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-13/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-13/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-13/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-13/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-13/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-13/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-13/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 38,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 30.0,
        "native_frames": 1200,
        "published_frames": 1200,
        "decode_ok": true,
        "sha256": "fd2b76def6534dec7070bdb4b4a181f9ebd6b2b14b27e653775bcc553e8ec767",
        "source_sha256": "57628073b4c5f789853928733509c15ee31b8b3938f93df5537d66b0ffdf6ca9",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-14",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/14",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "14",
      "seed": 100000,
      "source_episode_id": "14-click_alarmclock",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 64
      },
      "steps": 64,
      "simulation_time_s": 23.8160011311993,
      "wall_time_s": 12.765782033,
      "timing": {
        "wall_time_s": 12.765782033,
        "simulation_time_s": 23.8160011311993,
        "physics_steps": 5954,
        "wall_time_start_ns": 1793552027142594008,
        "wall_time_end_ns": 1793552039908379354,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Tap the center top button of the alarm-clock with beige screen",
      "native_instruction": "Tap the center top button of the alarm-clock with beige screen",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 64 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-14/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-14/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-14/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-14/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-14/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-14/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-14/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-14/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-14/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-14/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-14/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-14/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-14/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-14/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-14/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-14/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-14/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-14/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 2,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 1.625,
        "native_frames": 65,
        "published_frames": 65,
        "decode_ok": true,
        "sha256": "b41b6eb6f8ef7d03ba21939e09c931085e3edb031ff57c3c673d43264a1a9d6c",
        "source_sha256": "b72de613f6c095f4e2fe527068bacadc95d1835d8a129ee2c603dd2dbb38288f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-15",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/15",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "15",
      "seed": 100000,
      "source_episode_id": "15-click_bell",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 54
      },
      "steps": 54,
      "simulation_time_s": 23.880001134239137,
      "wall_time_s": 16.960771756,
      "timing": {
        "wall_time_s": 16.960771756,
        "simulation_time_s": 23.880001134239137,
        "physics_steps": 5970,
        "wall_time_start_ns": 1793552071906974863,
        "wall_time_end_ns": 1793552088867750109,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use the left arm to press the medium-sized bell's top center.",
      "native_instruction": "Use the left arm to press the medium-sized bell's top center.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 54 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-15/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-15/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-15/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-15/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-15/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-15/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-15/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-15/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-15/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-15/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-15/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-15/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-15/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-15/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-15/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-15/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-15/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-15/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 2,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 1.375,
        "native_frames": 55,
        "published_frames": 55,
        "decode_ok": true,
        "sha256": "9478b7209798f0f69bc3f02d278563a70783cab40c9870255635b7bf993eca47",
        "source_sha256": "66d49b2639d4aa2b95cd83219b005400d49fcdc079693c9b337a435912671079",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-16",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/16",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "16",
      "seed": 100006,
      "source_episode_id": "16-dump_bin_bigbin",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 289
      },
      "steps": 289,
      "simulation_time_s": 98.30400466918945,
      "wall_time_s": 40.956882474,
      "timing": {
        "wall_time_s": 40.956882474,
        "simulation_time_s": 98.30400466918945,
        "physics_steps": 24576,
        "wall_time_start_ns": 1793552130428310845,
        "wall_time_end_ns": 1793552171385196892,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
      "native_instruction": "Lift the little tabletrashbin with smooth texture and transfer the contents.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 289 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-16/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-16/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-16/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-16/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-16/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-16/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-16/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-16/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-16/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-16/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-16/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-16/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-16/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-16/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-16/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-16/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-16/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-16/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100006,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 10,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 7.25,
        "native_frames": 290,
        "published_frames": 290,
        "decode_ok": true,
        "sha256": "1a2763236e1ab5b4f0c5bda1260f3bd09983d48f7a0486068acb5218e347bd00",
        "source_sha256": "235f70b01013d7ad7f4f24aa0791dc92532d4396cffeb89a70e6493938926564",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-17",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/17",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "17",
      "seed": 100000,
      "source_episode_id": "17-grab_roller",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 75
      },
      "steps": 75,
      "simulation_time_s": 26.556001261342317,
      "wall_time_s": 16.819218105,
      "timing": {
        "wall_time_s": 16.819218105,
        "simulation_time_s": 26.556001261342317,
        "physics_steps": 6639,
        "wall_time_start_ns": 1793552141147301188,
        "wall_time_end_ns": 1793552157966522340,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use arms to grab the smooth wooden roller.",
      "native_instruction": "Use arms to grab the smooth wooden roller.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 75 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-17/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-17/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-17/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-17/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-17/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-17/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-17/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-17/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-17/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-17/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-17/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-17/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-17/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-17/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-17/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-17/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-17/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-17/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 1.9,
        "native_frames": 76,
        "published_frames": 76,
        "decode_ok": true,
        "sha256": "37a157ff2ab03bdb974ff13a82817a08791cab9f5da4ce78a981a2e8570942fe",
        "source_sha256": "3b42962eb386a1e09afcce72a3c6034a02f086d91e19671c527ae72fd4d66760",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-18",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/18",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "18",
      "seed": 100000,
      "source_episode_id": "18-handover_mic",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 180
      },
      "steps": 180,
      "simulation_time_s": 54.112002570182085,
      "wall_time_s": 26.516358336,
      "timing": {
        "wall_time_s": 26.516358336,
        "simulation_time_s": 54.112002570182085,
        "physics_steps": 13528,
        "wall_time_start_ns": 1793552194088846214,
        "wall_time_end_ns": 1793552220605207481,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use one arm to hold the microphone used for audio input and hand it over.",
      "native_instruction": "Use one arm to hold the microphone used for audio input and hand it over.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 180 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-18/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-18/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-18/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-18/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-18/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-18/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-18/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-18/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-18/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-18/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-18/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-18/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-18/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-18/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-18/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-18/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-18/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-18/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 6,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 4.525,
        "native_frames": 181,
        "published_frames": 181,
        "decode_ok": true,
        "sha256": "b088226e47e8ca210573ddd9882e2b1e4803d4cc8ef94093436f8e0d58c54b43",
        "source_sha256": "ae1e75578b7bf87bde53b70892afb2b7652e04db25d5f5faeb925e9bc054b28e",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-19",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/19",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "19",
      "seed": 100000,
      "source_episode_id": "19-move_can_pot",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 107.13200508849695,
      "wall_time_s": 48.017209828,
      "timing": {
        "wall_time_s": 48.017209828,
        "simulation_time_s": 107.13200508849695,
        "physics_steps": 26783,
        "wall_time_start_ns": 1793552209357332480,
        "wall_time_end_ns": 1793552257374545262,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
      "native_instruction": "Use the left arm to pick up the sauce can with red lid and move it beside the compact kitchenpot with lid-like surface",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-19/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-19/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-19/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-19/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-19/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-19/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-19/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-19/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-19/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-19/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-19/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-19/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-19/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-19/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-19/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-19/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-19/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-19/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "7537c3c2e3358f1fc01da7c2e13447d7b54316eb146c24ed8b3a1c4be6d1aad6",
        "source_sha256": "7525ece9997a63c7ff49e4e67e3f50add26c423ac3d5dc7cba29490d5d1bd5f0",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-20",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/20",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "20",
      "seed": 100001,
      "source_episode_id": "20-move_pillbottle_pad",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 110.70000525796786,
      "wall_time_s": 49.23542237,
      "timing": {
        "wall_time_s": 49.23542237,
        "simulation_time_s": 110.70000525796786,
        "physics_steps": 27675,
        "wall_time_start_ns": 1793552264611482878,
        "wall_time_end_ns": 1793552313846907824,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Grab the white cylindrical bottle and set it on the pad",
      "native_instruction": "Grab the white cylindrical bottle and set it on the pad",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-20/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-20/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-20/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-20/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-20/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-20/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-20/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-20/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-20/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-20/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-20/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-20/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-20/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-20/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-20/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-20/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-20/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-20/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "1defb09adab9825d6c3917471ee709d52bec87312e82756b6b74e4cd9349bdb8",
        "source_sha256": "40607239f6f381b093dc4c4b5d733a08bcaaf72d12e47286638157694b45d177",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-21",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/21",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "21",
      "seed": 100000,
      "source_episode_id": "21-move_playingcard_away",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 113.24400537880138,
      "wall_time_s": 43.386108159,
      "timing": {
        "wall_time_s": 43.386108159,
        "simulation_time_s": 113.24400537880138,
        "physics_steps": 28311,
        "wall_time_start_ns": 1793552287709289013,
        "wall_time_end_ns": 1793552331095401413,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use the right arm to grab the hard black box for card deck and relocate it.",
      "native_instruction": "Use the right arm to grab the hard black box for card deck and relocate it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-21/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-21/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-21/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-21/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-21/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-21/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-21/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-21/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-21/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-21/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-21/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-21/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-21/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-21/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-21/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-21/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-21/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-21/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "014422511a524aff7df908a937de80266716e7089e26a0167b86d7449e8e4a66",
        "source_sha256": "b57478a11dc6dbbf68f2c91865a759498bc7e139fb73fe2634848352594c1566",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-22",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/22",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "22",
      "seed": 100000,
      "source_episode_id": "22-move_stapler_pad",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 94.64400449534878,
      "wall_time_s": 46.038012018,
      "timing": {
        "wall_time_s": 46.038012018,
        "simulation_time_s": 94.64400449534878,
        "physics_steps": 23661,
        "wall_time_start_ns": 1793552346110921586,
        "wall_time_end_ns": 1793552392148936688,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use the left arm to move the plastic stapler to Yellow mat",
      "native_instruction": "Use the left arm to move the plastic stapler to Yellow mat",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-22/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-22/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-22/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-22/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-22/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-22/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-22/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-22/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-22/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-22/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-22/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-22/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-22/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-22/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-22/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-22/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-22/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-22/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "f7f00f8e510ce1f17ae9e27059fb089567121598a9b2182ff33717ec840fc2bb",
        "source_sha256": "4e3e4186a2980119fd7e002abc20664d11aa9d598b28d68171f6dd0802c6dc42",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-23",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/23",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "23",
      "seed": 100000,
      "source_episode_id": "23-open_laptop",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 138.35200657136738,
      "wall_time_s": 82.263339632,
      "timing": {
        "wall_time_s": 82.263339632,
        "simulation_time_s": 138.35200657136738,
        "physics_steps": 34588,
        "wall_time_start_ns": 1793552362750561871,
        "wall_time_end_ns": 1793552445013903698,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Open the flat orange laptop with one motion.",
      "native_instruction": "Open the flat orange laptop with one motion.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 700 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-23/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-23/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-23/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-23/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-23/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-23/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-23/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-23/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-23/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-23/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-23/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-23/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-23/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-23/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-23/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-23/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-23/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-23/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "30be2cf842bf940b77a01e2801a150913a53e65fc6158206b4937c1e819380bf",
        "source_sha256": "7be127e182b822f3af9bd4365e630359a0a549569c90d856041f9f9a4367d6e2",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-24",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/24",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "24",
      "seed": 100000,
      "source_episode_id": "24-open_microwave",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 1500
      },
      "steps": 1500,
      "simulation_time_s": 360.08401710307226,
      "wall_time_s": 143.692236395,
      "timing": {
        "wall_time_s": 143.692236395,
        "simulation_time_s": 360.08401710307226,
        "physics_steps": 90021,
        "wall_time_start_ns": 1793552429543373849,
        "wall_time_end_ns": 1793552573235612131,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
      "native_instruction": "Use the left arm to pull the handle of the medium gray microwave with top vents.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 1500 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-24/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-24/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-24/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-24/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-24/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-24/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-24/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-24/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-24/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-24/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-24/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-24/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-24/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-24/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-24/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-24/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-24/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-24/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 1500,
        "max_control_steps": 1500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 47,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 37.5,
        "native_frames": 1500,
        "published_frames": 1500,
        "decode_ok": true,
        "sha256": "3b41132d4949977b91ba52e3af61f13359709ab0de56839e877f3d9d5862e89e",
        "source_sha256": "293cc0c53f1ae876b662967b5b39782c3d0e27bf8f03056741a6336991620742",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768,
          800,
          832,
          864,
          896,
          928,
          960,
          992,
          1024,
          1056,
          1088,
          1120,
          1152,
          1184,
          1216,
          1248,
          1280,
          1312,
          1344,
          1376,
          1408,
          1440,
          1472
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-25",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/25",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "25",
      "seed": 100005,
      "source_episode_id": "25-pick_diverse_bottles",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 82.44000391568989,
      "wall_time_s": 46.256568292,
      "timing": {
        "wall_time_s": 46.256568292,
        "simulation_time_s": 82.44000391568989,
        "physics_steps": 20610,
        "wall_time_start_ns": 1793552494425322248,
        "wall_time_end_ns": 1793552540681893276,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
      "native_instruction": "Lift the bottle with white cap as the other arm lifts the glossy brown bottle.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-25/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-25/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-25/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-25/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-25/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-25/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-25/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-25/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-25/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-25/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-25/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-25/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-25/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-25/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-25/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-25/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-25/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-25/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100005,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "330ce9d72bf521d71f841daeefa72b5e4ed5e73a9b3e71d77488259616a2ddc6",
        "source_sha256": "912e7ed6009d39e5b29bd7afc37dcdeff3d998bd1faefbc512930c58af8b24c6",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-26",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/26",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "26",
      "seed": 100000,
      "source_episode_id": "26-pick_dual_bottles",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 81.65600387845188,
      "wall_time_s": 45.592737247,
      "timing": {
        "wall_time_s": 45.592737247,
        "simulation_time_s": 81.65600387845188,
        "physics_steps": 20414,
        "wall_time_start_ns": 1793552580281972898,
        "wall_time_end_ns": 1793552625874712388,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
      "native_instruction": "Hold the plastic soda bottle in one arm and the green bottle in the other.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-26/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-26/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-26/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-26/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-26/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-26/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-26/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-26/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-26/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-26/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-26/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-26/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-26/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-26/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-26/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-26/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-26/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-26/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "540cd4fc4df6dd0171eb9fc133692c4965b147cf7bc395958920bb27924e3ac7",
        "source_sha256": "37180e3b50db02d6553f985ef41db1f9377a17dca3fc58168521e1b92d413294",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-27",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/27",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "27",
      "seed": 100000,
      "source_episode_id": "27-place_a2b_left",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 124
      },
      "steps": 124,
      "simulation_time_s": 36.33600172586739,
      "wall_time_s": 21.860624988,
      "timing": {
        "wall_time_s": 21.860624988,
        "simulation_time_s": 36.33600172586739,
        "physics_steps": 9084,
        "wall_time_start_ns": 1793552603138017906,
        "wall_time_end_ns": 1793552624998646105,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
      "native_instruction": "Use the left arm to place the phone with rounded edges left of the hand-size bright blue stapler.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 124 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-27/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-27/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-27/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-27/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-27/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-27/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-27/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-27/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-27/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-27/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-27/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-27/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-27/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-27/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-27/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-27/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-27/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-27/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 3.125,
        "native_frames": 125,
        "published_frames": 125,
        "decode_ok": true,
        "sha256": "62ad24da95bd5995e2bf4256f829528cae1f30ac568d3290d54a8c75e0eb4a2b",
        "source_sha256": "bbb772c67814391277a8a1f38b70279d70cfda15508b3b684efb560ba6bef38f",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-28",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/28",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "28",
      "seed": 100000,
      "source_episode_id": "28-place_a2b_right",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 101.20400480693206,
      "wall_time_s": 45.117107136,
      "timing": {
        "wall_time_s": 45.117107136,
        "simulation_time_s": 101.20400480693206,
        "physics_steps": 25301,
        "wall_time_start_ns": 1793552655769322762,
        "wall_time_end_ns": 1793552700886433150,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
      "native_instruction": "Place the lightweight bell with smooth surface rightward of the light yellow cleaning soap using the left arm",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-28/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-28/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-28/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-28/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-28/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-28/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-28/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-28/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-28/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-28/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-28/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-28/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-28/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-28/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-28/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-28/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-28/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-28/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "bad83b0b4f5eab6f36550afe93ff30bcd97675bd284f32a58daec94fc3ddfe7f",
        "source_sha256": "17c6342cffad2f37b61b4c7f3371a7a876781617a26cadce7343680af8c70abc",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-29",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/29",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "29",
      "seed": 100000,
      "source_episode_id": "29-place_bread_basket",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 207.73600986693054,
      "wall_time_s": 90.165183701,
      "timing": {
        "wall_time_s": 90.165183701,
        "simulation_time_s": 207.73600986693054,
        "physics_steps": 51934,
        "wall_time_start_ns": 1793552661712194323,
        "wall_time_end_ns": 1793552751877381198,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
      "native_instruction": "Collect the square bread with light spots and the bread loaf with soft surface inside the beige breadbasket.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 700 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-29/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-29/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-29/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-29/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-29/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-29/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-29/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-29/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-29/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-29/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-29/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-29/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-29/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-29/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-29/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-29/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-29/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-29/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "7e61fcbccc1de53527a04ee27571aa83ff360a4b7efe3d9a969a9ff90bd8948c",
        "source_sha256": "11fed98fff8a80c010963476935982f9e425c47febfffef111eab1b9d9899bcd",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-30",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/30",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "30",
      "seed": 100001,
      "source_episode_id": "30-place_bread_skillet",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 97.32000462245196,
      "wall_time_s": 58.989428554,
      "timing": {
        "wall_time_s": 58.989428554,
        "simulation_time_s": 97.32000462245196,
        "physics_steps": 24330,
        "wall_time_start_ns": 1793552735723373440,
        "wall_time_end_ns": 1793552794712803935,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
      "native_instruction": "Use the left arm for the pan-shaped skillet with hole in handle and the other arm for the spongy golden brown bread.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 500 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-30/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-30/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-30/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-30/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-30/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-30/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-30/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-30/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-30/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-30/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-30/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-30/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-30/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-30/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-30/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-30/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-30/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-30/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "f07779f4abc413ca614c7e20593d6ad4bde252d53eb3888729e8303242ba1d87",
        "source_sha256": "7346a9fcfd007d2b0219cffed5db1c90c75aabdf23a673ffb57fd0c29ace4c99",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-31",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/31",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "31",
      "seed": 100000,
      "source_episode_id": "31-place_burger_fries",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 500
      },
      "steps": 500,
      "simulation_time_s": 152.2720072325319,
      "wall_time_s": 67.330002077,
      "timing": {
        "wall_time_s": 67.330002077,
        "simulation_time_s": 152.2720072325319,
        "physics_steps": 38068,
        "wall_time_start_ns": 1793552782856290846,
        "wall_time_end_ns": 1793552850186295788,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
      "native_instruction": "Place the hamburg with white sides and the Crispy potato fries with red box onto the smooth green tray.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 500 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-31/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-31/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-31/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-31/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-31/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-31/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-31/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-31/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-31/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-31/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-31/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-31/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-31/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-31/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-31/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-31/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-31/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-31/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 16,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 12.5,
        "native_frames": 500,
        "published_frames": 500,
        "decode_ok": true,
        "sha256": "8d08ebc0ec544f19edce99a1aa2124973bfea9f95791303b476f99d94208fdfa",
        "source_sha256": "33c9f68cfb25f80487af15322da73f928fda75282dcd45fc7940924d54cadc6b",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-32",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/32",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "32",
      "seed": 100001,
      "source_episode_id": "32-place_can_basket",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 190.31200903933495,
      "wall_time_s": 98.112168544,
      "timing": {
        "wall_time_s": 98.112168544,
        "simulation_time_s": 190.31200903933495,
        "physics_steps": 47578,
        "wall_time_start_ns": 1793552842244165248,
        "wall_time_end_ns": 1793552940356336359,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Grab the red can and put it in the medium-sized basket with perforated texture, then lift the medium-sized basket with perforated texture.",
      "native_instruction": "Grab the red can and put it in the medium-sized basket with perforated texture, then lift the medium-sized basket with perforated texture.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 700 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-32/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-32/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-32/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-32/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-32/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-32/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-32/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-32/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-32/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-32/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-32/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-32/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-32/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-32/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-32/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-32/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-32/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-32/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "ce8c8f3df313f51f6e5868842eaa78d642b1a9ac8d654664cdfe209e132b6f70",
        "source_sha256": "d23bdd508c122cff3354e5340f2759dbfb653aed9a9b1c2fc26000a0e34964c6",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-33",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/33",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "33",
      "seed": 100000,
      "source_episode_id": "33-place_cans_plasticbox",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 800
      },
      "steps": 800,
      "simulation_time_s": 229.8000109149143,
      "wall_time_s": 94.138474995,
      "timing": {
        "wall_time_s": 94.138474995,
        "simulation_time_s": 229.8000109149143,
        "physics_steps": 57450,
        "wall_time_start_ns": 1793552888799040109,
        "wall_time_end_ns": 1793552982937518229,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
      "native_instruction": "Lift the can with shiny smooth surface and the small red soda can, drop both into the portable light blue storage plasticbox.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 800 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-33/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-33/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-33/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-33/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-33/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-33/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-33/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-33/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-33/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-33/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-33/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-33/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-33/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-33/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-33/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-33/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-33/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-33/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 25,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 20.0,
        "native_frames": 800,
        "published_frames": 800,
        "decode_ok": true,
        "sha256": "c844a506667c44a5eff3b62c8ca02b347c935d43ff86aaf56a2b16cdba16919d",
        "source_sha256": "02ee2b403a5355652ee8bf5e81addeb7ce83a86c14a588838717cc507469d5bd",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672,
          704,
          736,
          768
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-34",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/34",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "34",
      "seed": 100000,
      "source_episode_id": "34-place_container_plate",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 129
      },
      "steps": 129,
      "simulation_time_s": 48.132002286147326,
      "wall_time_s": 26.706274076,
      "timing": {
        "wall_time_s": 26.706274076,
        "simulation_time_s": 48.132002286147326,
        "physics_steps": 12033,
        "wall_time_start_ns": 1793552970428610928,
        "wall_time_end_ns": 1793552997134892810,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Place the curved bowl with smooth texture on the gray plate for dining.",
      "native_instruction": "Place the curved bowl with smooth texture on the gray plate for dining.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 129 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-34/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-34/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-34/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-34/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-34/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-34/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-34/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-34/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-34/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-34/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-34/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-34/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-34/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-34/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-34/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-34/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-34/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-34/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 5,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 3.25,
        "native_frames": 130,
        "published_frames": 130,
        "decode_ok": true,
        "sha256": "24b6556b357fd6f647ce552d0243c11eabb3e5f2c5e7b9af642938f6994feb70",
        "source_sha256": "6692e44d84aeb3f147613d7701e71e67101d527f790ce92045d69ea26bb19802",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-35",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/35",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "35",
      "seed": 100000,
      "source_episode_id": "35-place_empty_cup",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 128
      },
      "steps": 128,
      "simulation_time_s": 46.01600218564272,
      "wall_time_s": 26.716689585,
      "timing": {
        "wall_time_s": 26.716689585,
        "simulation_time_s": 46.01600218564272,
        "physics_steps": 11504,
        "wall_time_start_ns": 1793553016098313475,
        "wall_time_end_ns": 1793553042815006551,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
      "native_instruction": "Set the handheld round blue cup on top of the disc-shaped light gray coaster.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 128 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-35/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-35/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-35/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-35/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-35/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-35/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-35/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-35/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-35/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-35/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-35/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-35/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-35/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-35/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-35/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-35/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-35/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-35/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 3.225,
        "native_frames": 129,
        "published_frames": 129,
        "decode_ok": true,
        "sha256": "a606c383e9eb86bf05ca0dbbb29a45c7280c7babb319013b71833737c8334881",
        "source_sha256": "242e9b88285f9b5ed3c59c1fa77302abf493f380c96293e29147df6b0cf68fdd",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-36",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/36",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "36",
      "seed": 100000,
      "source_episode_id": "36-place_fan",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 117.37200557487085,
      "wall_time_s": 52.362606151,
      "timing": {
        "wall_time_s": 52.362606151,
        "simulation_time_s": 117.37200557487085,
        "physics_steps": 29343,
        "wall_time_start_ns": 1793553026103312031,
        "wall_time_end_ns": 1793553078465920928,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
      "native_instruction": "Use the left arm to grab the plastic fan and place it on the Lime mat facing the robot.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-36/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-36/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-36/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-36/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-36/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-36/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-36/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-36/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-36/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-36/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-36/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-36/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-36/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-36/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-36/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-36/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-36/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-36/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "1121034c4ec30554e64412234e5e483f2efa018ec3a0bf273c1dae28c793437f",
        "source_sha256": "988e6bef955e3fae5561cfc25ba0822c85a443cca88fd296e5d89eac781fae98",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-37",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/37",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "37",
      "seed": 100000,
      "source_episode_id": "37-place_mouse_pad",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 111.20400528190657,
      "wall_time_s": 52.706938417,
      "timing": {
        "wall_time_s": 52.706938417,
        "simulation_time_s": 111.20400528190657,
        "physics_steps": 27801,
        "wall_time_start_ns": 1793553075450747420,
        "wall_time_end_ns": 1793553128157689005,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
      "native_instruction": "Lift the dark gray wireless mouse and drop it on the Yellow mat.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-37/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-37/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-37/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-37/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-37/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-37/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-37/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-37/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-37/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-37/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-37/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-37/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-37/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-37/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-37/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-37/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-37/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-37/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "22062f06ade3afeb8bc8c5cc1b741407de383e4aad656f5e4a78e762a203a0cf",
        "source_sha256": "8e4abb927b9070d78fd11354b293134fb4fac01e664af21d169585f716b8f942",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-38",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/38",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "38",
      "seed": 100000,
      "source_episode_id": "38-place_object_basket",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 700
      },
      "steps": 700,
      "simulation_time_s": 179.20400851173326,
      "wall_time_s": 103.140258386,
      "timing": {
        "wall_time_s": 103.140258386,
        "simulation_time_s": 179.20400851173326,
        "physics_steps": 44801,
        "wall_time_start_ns": 1793553110754957031,
        "wall_time_end_ns": 1793553213895218912,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
      "native_instruction": "Grab the toycar with rounded body, put it in the oval basket with medium storage space, move the oval basket with medium storage space.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 700 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-38/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-38/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-38/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-38/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-38/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-38/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-38/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-38/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-38/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-38/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-38/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-38/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-38/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-38/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-38/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-38/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-38/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-38/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 22,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 17.5,
        "native_frames": 700,
        "published_frames": 700,
        "decode_ok": true,
        "sha256": "f896b2ea37239e17fa29750328595610c9dcfb981abe6c9f050261b462fe6b7a",
        "source_sha256": "2f8ab8264c3e65ff2683d7780db4792ef5ff4a2682d63ee3917d33b1ad111637",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384,
          416,
          448,
          480,
          512,
          544,
          576,
          608,
          640,
          672
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-39",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/39",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "39",
      "seed": 100000,
      "source_episode_id": "39-place_object_scale",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 103.92000493593514,
      "wall_time_s": 52.971200962,
      "timing": {
        "wall_time_s": 52.971200962,
        "simulation_time_s": 103.92000493593514,
        "physics_steps": 25980,
        "wall_time_start_ns": 1793553161647997341,
        "wall_time_end_ns": 1793553214619201117,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
      "native_instruction": "With the left arm, grab the bell with raised round top and position it on the electronicscale with black digital screen",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-39/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-39/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-39/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-39/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-39/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-39/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-39/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-39/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-39/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-39/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-39/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-39/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-39/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-39/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-39/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-39/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-39/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-39/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "e263794cf7f76ca6aa21d846e023a4d90bebc95233c939c82064bd5ed9de716f",
        "source_sha256": "91e5c8da35ab6190efff0ea6f1d52a5482803269ac6f64b33b035c593141a491",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-40",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/40",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "40",
      "seed": 100001,
      "source_episode_id": "40-place_object_stand",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 118
      },
      "steps": 118,
      "simulation_time_s": 36.61600173916668,
      "wall_time_s": 21.80094009,
      "timing": {
        "wall_time_s": 21.80094009,
        "simulation_time_s": 36.61600173916668,
        "physics_steps": 9154,
        "wall_time_start_ns": 1793553248712868899,
        "wall_time_end_ns": 1793553270513811272,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Place the white stapler on the compact black hollow displaystand with the left arm.",
      "native_instruction": "Place the white stapler on the compact black hollow displaystand with the left arm.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 118 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-40/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-40/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-40/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-40/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-40/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-40/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-40/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-40/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-40/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-40/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-40/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-40/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-40/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-40/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-40/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-40/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-40/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-40/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 4,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 2.975,
        "native_frames": 119,
        "published_frames": 119,
        "decode_ok": true,
        "sha256": "7b22713e1338e6b9cd4e4c980ab5e940aa40e4bfa5c06de18608755701bb9360",
        "source_sha256": "0367c6bca5b1252e882dbc5b399fab51070d39eeeeedf00e5464444b6d54ef26",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-41",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/41",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "41",
      "seed": 100000,
      "source_episode_id": "41-place_phone_stand",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 117.63600558741018,
      "wall_time_s": 52.188806041,
      "timing": {
        "wall_time_s": 52.188806041,
        "simulation_time_s": 117.63600558741018,
        "physics_steps": 29409,
        "wall_time_start_ns": 1793553247160945144,
        "wall_time_end_ns": 1793553299349753933,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Set the phone with flat black screen on the angled brown phonestand.",
      "native_instruction": "Set the phone with flat black screen on the angled brown phonestand.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-41/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-41/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-41/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-41/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-41/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-41/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-41/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-41/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-41/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-41/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-41/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-41/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-41/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-41/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-41/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-41/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-41/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-41/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "69d549ec4a63ad3e728fb042ffdfc39ec31a8ab570569a81fca58c1cf62de5f5",
        "source_sha256": "0bf2eadb29e0f2ab934d5f784b37519732253b9a85919ef2399821d67b542c8a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-42",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/42",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "42",
      "seed": 100001,
      "source_episode_id": "42-place_shoe",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 129
      },
      "steps": 129,
      "simulation_time_s": 44.2200021003373,
      "wall_time_s": 24.130053601,
      "timing": {
        "wall_time_s": 24.130053601,
        "simulation_time_s": 44.2200021003373,
        "physics_steps": 11055,
        "wall_time_start_ns": 1793553304039321849,
        "wall_time_end_ns": 1793553328169380075,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Pick up the medium adult shoe from the table and set it on the mat",
      "native_instruction": "Pick up the medium adult shoe from the table and set it on the mat",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 129 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-42/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-42/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-42/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-42/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-42/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-42/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-42/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-42/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-42/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-42/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-42/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-42/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-42/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-42/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-42/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-42/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-42/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-42/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 5,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 3.25,
        "native_frames": 130,
        "published_frames": 130,
        "decode_ok": true,
        "sha256": "40757a8a59316bcf7c6166df83639316f549860ec03ec72431cb66d2f692c3bf",
        "source_sha256": "baef440a1b535d148be279a7e969641f214163891b2566ebca1eb1f001096ce5",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-43",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/43",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "43",
      "seed": 100000,
      "source_episode_id": "43-press_stapler",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 70.34000334097072,
      "wall_time_s": 44.325573874,
      "timing": {
        "wall_time_s": 44.325573874,
        "simulation_time_s": 70.34000334097072,
        "physics_steps": 17585,
        "wall_time_start_ns": 1793553331549736128,
        "wall_time_end_ns": 1793553375875312793,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Press the plastic stapler firmly using your arm.",
      "native_instruction": "Press the plastic stapler firmly using your arm.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-43/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-43/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-43/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-43/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-43/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-43/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-43/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-43/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-43/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-43/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-43/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-43/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-43/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-43/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-43/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-43/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-43/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-43/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "dda36f860f98bd68d90d2edb2e9b1999290d5e84efc411c4fd06da1bea9d2b41",
        "source_sha256": "398ad8e3f66c86c89a093dfc718768c8e3e871429107ee5d68885ea43b57454a",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-44",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/44",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "44",
      "seed": 100000,
      "source_episode_id": "44-rotate_qrcode",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 113.68800539989024,
      "wall_time_s": 51.584260198,
      "timing": {
        "wall_time_s": 51.584260198,
        "simulation_time_s": 113.68800539989024,
        "physics_steps": 28422,
        "wall_time_start_ns": 1793553357938393977,
        "wall_time_end_ns": 1793553409522656566,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
      "native_instruction": "Pick up the QR code payment sign, rotate its QR code to face you, then set it down.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-44/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-44/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-44/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-44/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-44/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-44/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-44/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-44/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-44/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-44/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-44/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-44/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-44/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-44/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-44/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-44/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-44/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-44/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "d3a56f1c053f64fa16e7edc652b8b964e813d1947c94b497a28b1f328a83f76c",
        "source_sha256": "75b1a5255ed72a333cf496a66b99c5b50edb57f04702add6def0ac55d39de361",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-45",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/45",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "45",
      "seed": 100001,
      "source_episode_id": "45-shake_bottle",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 61
      },
      "steps": 61,
      "simulation_time_s": 21.76000103354454,
      "wall_time_s": 16.583793808,
      "timing": {
        "wall_time_s": 16.583793808,
        "simulation_time_s": 21.76000103354454,
        "physics_steps": 5440,
        "wall_time_start_ns": 1793553410157522870,
        "wall_time_end_ns": 1793553426741321302,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Pick up the smooth bottle with blue text and shake it.",
      "native_instruction": "Pick up the smooth bottle with blue text and shake it.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 61 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-45/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-45/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-45/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-45/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-45/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-45/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-45/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-45/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-45/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-45/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-45/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-45/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-45/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-45/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-45/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-45/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-45/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-45/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 2,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 1.55,
        "native_frames": 62,
        "published_frames": 62,
        "decode_ok": true,
        "sha256": "a761a0fa6865f6d269a7a7c1c2e669831d4c9cc3ee1cfc1b0a8a1e618daeb1d6",
        "source_sha256": "47ef5db6adbcddcd4e7d268d671c329a618a7cff28245490987d011e13f7bf7c",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-46",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/46",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "46",
      "seed": 100001,
      "source_episode_id": "46-shake_bottle_horizontally",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 73
      },
      "steps": 73,
      "simulation_time_s": 25.132001193705946,
      "wall_time_s": 18.643482662,
      "timing": {
        "wall_time_s": 18.643482662,
        "simulation_time_s": 25.132001193705946,
        "physics_steps": 6283,
        "wall_time_start_ns": 1793553448493388360,
        "wall_time_end_ns": 1793553467136875339,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
      "native_instruction": "Grab the smooth bottle with blue text with the left arm, shake horizontally.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 73 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-46/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-46/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-46/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-46/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-46/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-46/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-46/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-46/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-46/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-46/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-46/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-46/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-46/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-46/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-46/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-46/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-46/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-46/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100001,
        "episodes": 1,
        "max_steps": 700,
        "max_control_steps": 700,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 3,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 1.85,
        "native_frames": 74,
        "published_frames": 74,
        "decode_ok": true,
        "sha256": "83c75809accedb065ac59288ad379ba5c926f81291b91e47e7346ee66b986dad",
        "source_sha256": "f525e86924f88c150c4e8c506f8fb0726cda5204a4bb760c148383e77bd9a51e",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-47",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/47",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "47",
      "seed": 100000,
      "source_episode_id": "47-stack_blocks_two",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 274
      },
      "steps": 274,
      "simulation_time_s": 83.66400397382677,
      "wall_time_s": 38.375296278,
      "timing": {
        "wall_time_s": 38.375296278,
        "simulation_time_s": 83.66400397382677,
        "physics_steps": 20916,
        "wall_time_start_ns": 1793553458681708977,
        "wall_time_end_ns": 1793553497057008860,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Move red block to the center, then put green block on top.",
      "native_instruction": "Move red block to the center, then put green block on top.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 274 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-47/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-47/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-47/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-47/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-47/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-47/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-47/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-47/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-47/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-47/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-47/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-47/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-47/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-47/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-47/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-47/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-47/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-47/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 9,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 6.875,
        "native_frames": 275,
        "published_frames": 275,
        "decode_ok": true,
        "sha256": "b5bd9704a9c8e2b2c32f021544c15273ff182c941fd5021db9de831f6649c6b4",
        "source_sha256": "d570f006cdf942da87b8f0fe5d9ea24c8233fee1db37d6e706364f50035540f4",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-48",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/48",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "48",
      "seed": 100000,
      "source_episode_id": "48-stack_bowls_two",
      "status": "archived_native_instruction_protocol",
      "success": true,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": true,
        "evidence_valid": true,
        "termination": "success",
        "steps": 316
      },
      "steps": 316,
      "simulation_time_s": 92.40400438895449,
      "wall_time_s": 41.106791051,
      "timing": {
        "wall_time_s": 41.106791051,
        "simulation_time_s": 92.40400438895449,
        "physics_steps": 23101,
        "wall_time_start_ns": 1793553500960187693,
        "wall_time_end_ns": 1793553542066983903,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
      "native_instruction": "Place the light blue bowl with darkened rim carefully over the light blue bowl with darkened rim to stack them.",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native success after 316 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-48/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-48/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-48/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-48/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-48/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-48/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-48/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-48/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-48/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-48/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-48/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-48/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-48/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-48/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-48/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-48/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-48/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-48/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 10,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 7.925,
        "native_frames": 317,
        "published_frames": 317,
        "decode_ok": true,
        "sha256": "063f2aa98fe2d55a89ff5e298f27158d7f7382532428a46053b0fa7e545db891",
        "source_sha256": "148ffc8c009ed0b1e50a4607880bfc7a10a70a14f62f937a85045b4d20bbf2c1",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": true,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-49",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/49",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "49",
      "seed": 100003,
      "source_episode_id": "49-stamp_seal",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 120.77200573636219,
      "wall_time_s": 54.047483176,
      "timing": {
        "wall_time_s": 54.047483176,
        "simulation_time_s": 120.77200573636219,
        "physics_steps": 30193,
        "wall_time_start_ns": 1793553537475230526,
        "wall_time_end_ns": 1793553591522717353,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Grab the block-shaped seal with angled edges, press onto Blue",
      "native_instruction": "Grab the block-shaped seal with angled edges, press onto Blue",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-49/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-49/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-49/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-49/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-49/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-49/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-49/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-49/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-49/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-49/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-49/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-49/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-49/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-49/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-49/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-49/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-49/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-49/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100003,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "39466e3e71cdd9ad86971c7f30ac62786b292609a533ea355cbcc078bf774fea",
        "source_sha256": "0a8cc2e8471590e89aa52b55a528cd6fa173835d1ac84e922fd810f5dc05959e",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "hard-clean-50",
      "variant": "standalone-vlact-clean",
      "task_key": "task03/50",
      "family": "task03",
      "source_family": "task03",
      "provider": "robotwin-official",
      "slot": "50",
      "seed": 100000,
      "source_episode_id": "50-turn_switch",
      "status": "archived_native_instruction_protocol",
      "success": false,
      "valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "success": false,
        "evidence_valid": true,
        "termination": "step_limit",
        "steps": 400
      },
      "steps": 400,
      "simulation_time_s": 93.06400442030281,
      "wall_time_s": 48.117566427,
      "timing": {
        "wall_time_s": 48.117566427,
        "simulation_time_s": 93.06400442030281,
        "physics_steps": 23266,
        "wall_time_start_ns": 1793553575420586481,
        "wall_time_end_ns": 1793553623538155352,
        "timing_scope": "after native setup to last action completion",
        "wall_time_source": "monotonic clock",
        "simulation_time_source": "sum of timestep for completed scene.step calls"
      },
      "model": "VLAct Qwen3-VL-OFT",
      "effort": "50k · official",
      "harness": "standalone",
      "instruction": "Press the flat square gray switch using the left arm",
      "native_instruction": "Press the flat square gray switch using the left arm",
      "instruction_policy": "original_native",
      "usage": {
        "schema": "rlebench/token-usage/1",
        "source": "no-language-agent",
        "usage_source": "no-language-agent",
        "response_count": 0,
        "input_tokens": 0,
        "output_tokens": 0,
        "cached_input_tokens": 0,
        "cache_write_input_tokens": 0,
        "fresh_input_tokens": 0,
        "reasoning_output_tokens": 0,
        "audit_complete": true,
        "response_ids_complete": true
      },
      "call_activity": {
        "model_tool_calls": 0
      },
      "analysis": {
        "observed": "Native failure after 400 official actions. One valid episode with the native unseen instruction; no valid-outcome retry."
      },
      "links": {
        "video": "experiments/robotwin-hard-20261101/hard-clean-50/presentation/video.mp4",
        "poster": "experiments/robotwin-hard-20261101/hard-clean-50/presentation/poster.jpg",
        "terminal_frame": "experiments/robotwin-hard-20261101/hard-clean-50/presentation/terminal.jpg",
        "transcript": "experiments/robotwin-hard-20261101/hard-clean-50/transcript.json",
        "usage": "experiments/robotwin-hard-20261101/hard-clean-50/usage.json",
        "analysis": "experiments/robotwin-hard-20261101/hard-clean-50/analysis.json",
        "protocol": "experiments/robotwin-hard-20261101/hard-clean-50/protocol.json",
        "native_goal": "experiments/robotwin-hard-20261101/hard-clean-50/instruction.json",
        "verdict": "experiments/robotwin-hard-20261101/hard-clean-50/episode.json",
        "media_check": "experiments/robotwin-hard-20261101/hard-clean-50/presentation/media-validation.json",
        "recording_manifest": "experiments/robotwin-hard-20261101/hard-clean-50/presentation/media-validation.json",
        "tool_calls": "experiments/robotwin-hard-20261101/hard-clean-50/tool-calls.json",
        "artifact_index": "experiments/robotwin-hard-20261101/hard-clean-50/artifact-index.json",
        "vla_io": "experiments/robotwin-hard-20261101/hard-clean-50/vla-io.json",
        "evidence_archive": "experiments/robotwin-hard-20261101/hard-clean-50/evidence.zip",
        "timing": "experiments/robotwin-hard-20261101/hard-clean-50/timing.json",
        "provenance": "research/robotwin-hard-20261101/source-identities.json",
        "native_video": "experiments/robotwin-hard-20261101/hard-clean-50/native-head.mp4",
        "policy_trace": "experiments/robotwin-hard-20261101/hard-clean-50/policy/events.jsonl"
      },
      "resources": [],
      "session_counts": {
        "visible_events": 0,
        "observed_images": 0
      },
      "measured_protocol": {
        "seed": 100000,
        "episodes": 1,
        "max_steps": 400,
        "max_control_steps": 400,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_type": "unseen",
        "provider": "robotwin-official",
        "control_frequency_hz": null,
        "mode": "native_trajectory",
        "admission": "official expert check"
      },
      "vla_calls": 13,
      "media": {
        "speed": 4,
        "speed_basis": "official 10 FPS action-index recording, not physical simulation time",
        "playback_kind": "official_action_index",
        "source_fps": 10,
        "output_fps": 40,
        "width": 2880,
        "height": 720,
        "duration_s": 10.0,
        "native_frames": 400,
        "published_frames": 400,
        "decode_ok": true,
        "sha256": "f2eefd8d980569404246ee18cba8d3ce88da57b38c39d43c9855dd5812b1e4ac",
        "source_sha256": "a4c0803c228db51c017be70b78ac2d500468fd28b25e24b32b4720b0269a675c",
        "head_frames_preserved": true,
        "terminal_head_frame_recorded": false,
        "wrist_sampling": "Inference observations only, held from their recorded native_step until the next inference. Final wrist observations can precede termination.",
        "resolution_kind": "Native 320 x 240 RGB per view, upscaled to 960 x 720 for the reference layout",
        "camera_order": [
          "head",
          "left wrist",
          "right wrist"
        ],
        "native_step_by_inference": [
          0,
          32,
          64,
          96,
          128,
          160,
          192,
          224,
          256,
          288,
          320,
          352,
          384
        ],
        "last_frame_note": "Successful runs include a post-action head frame; action-limit failures end on the last pre-action head frame. No new terminal frames were synthesized."
      },
      "archived_reason": "Earlier native-instruction run of the same checkpoint (16/50; per-run expert checks, its own seeds on 4 tasks). Replaced in current results by the 2026-11-15 baseline on the campaign scenes; original evidence retained."
    },
    {
      "id": "native-full-kinex-tcp-posestate-04-20261113",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/04",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "04",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_superseded_by_onboard_rerun_20261115",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 10.200000000000001,
        "step_limit": 500,
        "steps": 255,
        "success": false,
        "valid": true
      },
      "steps": 255,
      "simulation_time_s": 10.200000000000001,
      "wall_time_s": 598.38022,
      "timing": {
        "simulation_time_s": 10.200000000000001,
        "wall_time_s": 598.38022,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "native_instruction": "Lift both the cube tea box with shiny embossed patterns and the trigger scanner at once before scanning the cube tea box with shiny embossed patterns.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9681435725696756,
        "cache_read_input_tokens": 881152,
        "cache_reported_input_tokens": 910146,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 910146,
        "cached_input_tokens": 881152,
        "cost_usd": null,
        "fresh_input_tokens": 28994,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 910146,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 881152,
        "known_fresh_input_tokens": 28994,
        "known_input_tokens": 910146,
        "known_output_tokens": 6900,
        "known_reasoning_output_tokens": 2120,
        "output_tokens": 6900,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2120,
        "reasoning_reported_output_tokens": 6900,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 33,
          "cache_write_input_tokens": 33,
          "cache_write_reported_input_tokens": 33,
          "cached_input_tokens": 33,
          "fresh_input_tokens": 33,
          "input_tokens": 33,
          "output_tokens": 33,
          "reasoning_output_tokens": 33,
          "reasoning_reported_output_tokens": 33
        },
        "request_attempts": 33,
        "response_count": 33,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 917046,
        "uncached_input_tokens": 28994,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11828,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11868,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12001,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12345,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18051,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 19313,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 19883,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19840,
            "input_tokens": 20113,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20303,
            "output_tokens": 199,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20352,
            "input_tokens": 22906,
            "output_tokens": 363,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23168,
            "input_tokens": 23664,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23552,
            "input_tokens": 24072,
            "output_tokens": 351,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24320,
            "input_tokens": 26114,
            "output_tokens": 1142,
            "reasoning_output_tokens": 360
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27136,
            "input_tokens": 27277,
            "output_tokens": 129,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27512,
            "output_tokens": 141,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28273,
            "output_tokens": 68,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28368,
            "output_tokens": 141,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 29206,
            "output_tokens": 294,
            "reasoning_output_tokens": 157
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30195,
            "output_tokens": 64,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30633,
            "output_tokens": 153,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31482,
            "output_tokens": 111,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31488,
            "input_tokens": 33133,
            "output_tokens": 119,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33315,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33675,
            "output_tokens": 505,
            "reasoning_output_tokens": 419
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34252,
            "output_tokens": 186,
            "reasoning_output_tokens": 69
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34957,
            "output_tokens": 337,
            "reasoning_output_tokens": 85
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35744,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36620,
            "output_tokens": 263,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36736,
            "input_tokens": 37256,
            "output_tokens": 236,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37376,
            "input_tokens": 37649,
            "output_tokens": 298,
            "reasoning_output_tokens": 199
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37760,
            "input_tokens": 38484,
            "output_tokens": 463,
            "reasoning_output_tokens": 370
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38784,
            "input_tokens": 39477,
            "output_tokens": 678,
            "reasoning_output_tokens": 203
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40177,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 43,
        "model_tool_calls_by_name": {
          "ls": 1,
          "read": 18,
          "bash": 1,
          "exec": 16,
          "devices": 3,
          "device": 1,
          "write": 2,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 255 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; full 50-task suite.",
        "execution": "Kinex + progVLA-jy 2026-11-13 result, replaced by the 2026-11-15 onboard rerun of tasks 01-05."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/transcript.json",
        "usage": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/usage.json",
        "analysis": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/analysis.json",
        "protocol": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/instruction.json",
        "verdict": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/episode.json",
        "media_check": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/evidence.zip",
        "timing": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/timing.json",
        "native_session": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/session.jsonl",
        "trace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/scan-object.md",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/workspace/.kinex/memos/scan-object.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/dual_tcp.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-04-20261113/workspace/.kinex/tools/dual_tcp.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 79,
        "observed_images": 11
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 500,
        "max_control_steps": 500,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-progress-off-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 103,
        "published_frames": 103,
        "decode_ok": true,
        "duration_s": 2.575,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "608b30e61ebb4e11d1314a3167a0113e2929446bfc316853e6cda9b9b65d3dc7",
        "source_sha256": {
          "third_person": "1a013d72e796831050cfac4105d5e45e8db7ec9c6ba93b110f3f7dc77e38231b",
          "left_wrist": "56567ddeddaeb75e45280bd08252bc9f495918eaaa2ed380061b9c92c7490225",
          "right_wrist": "f59ea5c4cb2275058382e2b8b6afa42dfb208d8c922e789aa393beef5b9b4592"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "7816fba8b484",
          "task": "robotwin/scan-object-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 104,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 103,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10200000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 104,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 103,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10200000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 104,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 103,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 10200000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-progress-off-20261113",
      "vla_model": "progVLA-jy",
      "vla_used": false,
      "superseded_by": "onboard-rerun-kinex-tcp-posestate-04-20261115",
      "archived_at": "2026-11-16T02:53:53.474121+00:00",
      "source_dataset_revision": "8cdd263ca04b3a249f3c8a4f66a653e2a890c8fb"
    },
    {
      "id": "native-full-kinex-tcp-posestate-01-20261113",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/01",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "01",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_superseded_by_onboard_rerun_20261115",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 27.44,
        "step_limit": 800,
        "steps": 686,
        "success": true,
        "valid": true
      },
      "steps": 686,
      "simulation_time_s": 27.44,
      "wall_time_s": 2061.529521,
      "timing": {
        "simulation_time_s": 27.44,
        "wall_time_s": 2061.529521,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "native_instruction": "Grab the red block with the left arm, transfer it to the right arm, and set it on the blue pad.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9805051259616652,
        "cache_read_input_tokens": 2575232,
        "cache_reported_input_tokens": 2626434,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2626434,
        "cached_input_tokens": 2575232,
        "cost_usd": null,
        "fresh_input_tokens": 51202,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2626434,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2575232,
        "known_fresh_input_tokens": 51202,
        "known_input_tokens": 2626434,
        "known_output_tokens": 14595,
        "known_reasoning_output_tokens": 6491,
        "output_tokens": 14595,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 6491,
        "reasoning_reported_output_tokens": 14595,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 65,
          "cache_write_input_tokens": 65,
          "cache_write_reported_input_tokens": 65,
          "cached_input_tokens": 65,
          "fresh_input_tokens": 65,
          "input_tokens": 65,
          "output_tokens": 65,
          "reasoning_output_tokens": 65,
          "reasoning_reported_output_tokens": 65
        },
        "request_attempts": 65,
        "response_count": 65,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2641029,
        "uncached_input_tokens": 51202,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11824,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11955,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12003,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12296,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12447,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 18008,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 19270,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19200,
            "input_tokens": 19840,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19712,
            "input_tokens": 20070,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20261,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20096,
            "input_tokens": 20513,
            "output_tokens": 153,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20480,
            "input_tokens": 22536,
            "output_tokens": 478,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22912,
            "input_tokens": 25033,
            "output_tokens": 509,
            "reasoning_output_tokens": 92
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26116,
            "output_tokens": 272,
            "reasoning_output_tokens": 230
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26240,
            "input_tokens": 27567,
            "output_tokens": 1088,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 28676,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28544,
            "input_tokens": 29404,
            "output_tokens": 135,
            "reasoning_output_tokens": 52
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 30191,
            "output_tokens": 52,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30877,
            "output_tokens": 124,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30976,
            "input_tokens": 31648,
            "output_tokens": 330,
            "reasoning_output_tokens": 215
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32634,
            "output_tokens": 224,
            "reasoning_output_tokens": 148
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33489,
            "output_tokens": 82,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33633,
            "output_tokens": 393,
            "reasoning_output_tokens": 312
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 35566,
            "output_tokens": 197,
            "reasoning_output_tokens": 158
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 35826,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35712,
            "input_tokens": 36018,
            "output_tokens": 601,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36480,
            "input_tokens": 37291,
            "output_tokens": 82,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37865,
            "output_tokens": 152,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38712,
            "output_tokens": 473,
            "reasoning_output_tokens": 180
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39350,
            "output_tokens": 302,
            "reasoning_output_tokens": 9
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40120,
            "output_tokens": 332,
            "reasoning_output_tokens": 222
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40320,
            "input_tokens": 40944,
            "output_tokens": 217,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41188,
            "output_tokens": 107,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41926,
            "output_tokens": 282,
            "reasoning_output_tokens": 203
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42906,
            "output_tokens": 143,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42880,
            "input_tokens": 43699,
            "output_tokens": 162,
            "reasoning_output_tokens": 80
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44558,
            "output_tokens": 284,
            "reasoning_output_tokens": 173
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45019,
            "output_tokens": 110,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44928,
            "input_tokens": 45779,
            "output_tokens": 147,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45696,
            "input_tokens": 46557,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46592,
            "input_tokens": 47221,
            "output_tokens": 106,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47937,
            "output_tokens": 112,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47872,
            "input_tokens": 48678,
            "output_tokens": 234,
            "reasoning_output_tokens": 120
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48768,
            "input_tokens": 49561,
            "output_tokens": 135,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49536,
            "input_tokens": 49758,
            "output_tokens": 1023,
            "reasoning_output_tokens": 928
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49664,
            "input_tokens": 51404,
            "output_tokens": 533,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51840,
            "input_tokens": 51957,
            "output_tokens": 31,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51840,
            "input_tokens": 52296,
            "output_tokens": 215,
            "reasoning_output_tokens": 130
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52352,
            "input_tokens": 53136,
            "output_tokens": 87,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53120,
            "input_tokens": 53828,
            "output_tokens": 243,
            "reasoning_output_tokens": 125
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53888,
            "input_tokens": 54770,
            "output_tokens": 145,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54784,
            "input_tokens": 55538,
            "output_tokens": 49,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55424,
            "input_tokens": 55896,
            "output_tokens": 429,
            "reasoning_output_tokens": 347
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56192,
            "input_tokens": 56948,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56960,
            "input_tokens": 57287,
            "output_tokens": 494,
            "reasoning_output_tokens": 409
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57728,
            "input_tokens": 58409,
            "output_tokens": 113,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 58368,
            "input_tokens": 59129,
            "output_tokens": 492,
            "reasoning_output_tokens": 410
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 59520,
            "input_tokens": 60245,
            "output_tokens": 71,
            "reasoning_output_tokens": 37
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60160,
            "input_tokens": 60624,
            "output_tokens": 426,
            "reasoning_output_tokens": 89
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 60928,
            "input_tokens": 61313,
            "output_tokens": 305,
            "reasoning_output_tokens": 194
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 61440,
            "input_tokens": 62250,
            "output_tokens": 161,
            "reasoning_output_tokens": 79
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62208,
            "input_tokens": 63047,
            "output_tokens": 71,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 62976,
            "input_tokens": 63260,
            "output_tokens": 114,
            "reasoning_output_tokens": 62
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 63232,
            "input_tokens": 63770,
            "output_tokens": 764,
            "reasoning_output_tokens": 133
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 64384,
            "input_tokens": 64557,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 73,
        "model_tool_calls_by_name": {
          "read": 18,
          "ls": 1,
          "bash": 1,
          "exec": 42,
          "devices": 3,
          "device": 2,
          "write": 3,
          "device_stop": 2,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 686 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; full 50-task suite.",
        "execution": "Kinex + progVLA-jy 2026-11-13 result, replaced by the 2026-11-15 onboard rerun of tasks 01-05."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/transcript.json",
        "usage": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/usage.json",
        "analysis": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/analysis.json",
        "protocol": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/instruction.json",
        "verdict": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/episode.json",
        "media_check": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/evidence.zip",
        "timing": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/timing.json",
        "native_session": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/session.jsonl",
        "trace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/handover-block.md",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/workspace/.kinex/memos/handover-block.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/color_geometry.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-01-20261113/workspace/.kinex/tools/color_geometry.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 141,
        "observed_images": 26
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 800,
        "max_control_steps": 800,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-progress-off-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 276,
        "published_frames": 276,
        "decode_ok": true,
        "duration_s": 6.9,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "1c28f82e5c03fe9277d2dd5b19d865432e2f2a2e250b0dbac1e05d3d0866c0b8",
        "source_sha256": {
          "third_person": "00d37b77ba48977efde444cb7c1d85b60fffbca3719067e1bbee9b7b90e46b65",
          "left_wrist": "54800e7dbbf658d7f83da2ab44691dbee403193181b05ce2aceb9480a1efd291",
          "right_wrist": "0115759df2e1b6e433c60580a5abd1ef27ae1ba4c666f63f3b0da9887e869dce"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "503a6d43ed18",
          "task": "robotwin/handover-block-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 276,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 276,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27440000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 276,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 276,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27440000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 276,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 276,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 27440000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-progress-off-20261113",
      "vla_model": "progVLA-jy",
      "vla_used": false,
      "superseded_by": "onboard-rerun-kinex-tcp-posestate-01-20261115",
      "archived_at": "2026-11-16T02:53:53.474121+00:00",
      "source_dataset_revision": "8cdd263ca04b3a249f3c8a4f66a653e2a890c8fb"
    },
    {
      "id": "native-full-kinex-tcp-posestate-02-20261113",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/02",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "02",
      "seed": 1,
      "source_episode_id": "episode_01",
      "status": "archived_superseded_by_onboard_rerun_20261115",
      "success": false,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 1
        },
        "native_reward": 0.0,
        "planned_episodes": 1,
        "reason": "stopped",
        "simulation_time_s": 21.68,
        "step_limit": 600,
        "steps": 542,
        "success": false,
        "valid": true
      },
      "steps": 542,
      "simulation_time_s": 21.68,
      "wall_time_s": 1609.381044,
      "timing": {
        "simulation_time_s": 21.68,
        "wall_time_s": 1609.381044,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "native_instruction": "Pick up two the red shoe, tips left, into the medium cardboard shoe-box.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9689599148701675,
        "cache_read_input_tokens": 1223808,
        "cache_reported_input_tokens": 1263012,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1263012,
        "cached_input_tokens": 1223808,
        "cost_usd": null,
        "fresh_input_tokens": 39204,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1263012,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1223808,
        "known_fresh_input_tokens": 39204,
        "known_input_tokens": 1263012,
        "known_output_tokens": 8067,
        "known_reasoning_output_tokens": 3009,
        "output_tokens": 8067,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 3009,
        "reasoning_reported_output_tokens": 8067,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 43,
          "cache_write_input_tokens": 43,
          "cache_write_reported_input_tokens": 43,
          "cached_input_tokens": 43,
          "fresh_input_tokens": 43,
          "input_tokens": 43,
          "output_tokens": 43,
          "reasoning_output_tokens": 43,
          "reasoning_reported_output_tokens": 43
        },
        "request_attempts": 43,
        "response_count": 43,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1271079,
        "uncached_input_tokens": 39204,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11818,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11951,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11999,
            "output_tokens": 123,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12032,
            "input_tokens": 12333,
            "output_tokens": 41,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12288,
            "input_tokens": 12480,
            "output_tokens": 227,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 18041,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 19304,
            "output_tokens": 170,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19328,
            "input_tokens": 20101,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19968,
            "input_tokens": 20290,
            "output_tokens": 128,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 20224,
            "input_tokens": 22129,
            "output_tokens": 226,
            "reasoning_output_tokens": 74
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22750,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22656,
            "input_tokens": 23158,
            "output_tokens": 377,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 23424,
            "input_tokens": 24952,
            "output_tokens": 249,
            "reasoning_output_tokens": 207
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25088,
            "input_tokens": 26380,
            "output_tokens": 1045,
            "reasoning_output_tokens": 190
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27446,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 27903,
            "output_tokens": 139,
            "reasoning_output_tokens": 63
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27904,
            "input_tokens": 28412,
            "output_tokens": 46,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 28834,
            "output_tokens": 117,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29322,
            "output_tokens": 338,
            "reasoning_output_tokens": 128
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30072,
            "output_tokens": 242,
            "reasoning_output_tokens": 168
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30691,
            "output_tokens": 100,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31167,
            "output_tokens": 90,
            "reasoning_output_tokens": 49
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31104,
            "input_tokens": 31279,
            "output_tokens": 143,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31796,
            "output_tokens": 71,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 32248,
            "output_tokens": 634,
            "reasoning_output_tokens": 516
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32768,
            "input_tokens": 33255,
            "output_tokens": 187,
            "reasoning_output_tokens": 110
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33280,
            "input_tokens": 33816,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34238,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34686,
            "output_tokens": 189,
            "reasoning_output_tokens": 110
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35252,
            "output_tokens": 267,
            "reasoning_output_tokens": 187
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35456,
            "input_tokens": 35902,
            "output_tokens": 45,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35840,
            "input_tokens": 36324,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 36780,
            "output_tokens": 666,
            "reasoning_output_tokens": 520
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37822,
            "output_tokens": 183,
            "reasoning_output_tokens": 98
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37888,
            "input_tokens": 38382,
            "output_tokens": 56,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38272,
            "input_tokens": 38460,
            "output_tokens": 80,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38917,
            "output_tokens": 231,
            "reasoning_output_tokens": 146
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39040,
            "input_tokens": 39527,
            "output_tokens": 66,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 39615,
            "output_tokens": 97,
            "reasoning_output_tokens": 25
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39680,
            "input_tokens": 40087,
            "output_tokens": 139,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40064,
            "input_tokens": 40602,
            "output_tokens": 327,
            "reasoning_output_tokens": 142
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 40956,
            "output_tokens": 556,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41535,
            "output_tokens": 26,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 60,
        "model_tool_calls_by_name": {
          "read": 15,
          "ls": 4,
          "find": 1,
          "bash": 1,
          "exec": 26,
          "device": 7,
          "write": 2,
          "device_stop": 3,
          "edit": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native unsuccessful outcome after 542 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; full 50-task suite.",
        "execution": "Kinex + progVLA-jy 2026-11-13 result, replaced by the 2026-11-15 onboard rerun of tasks 01-05."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/transcript.json",
        "usage": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/usage.json",
        "analysis": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/analysis.json",
        "protocol": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/instruction.json",
        "verdict": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/episode.json",
        "media_check": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/evidence.zip",
        "timing": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/timing.json",
        "native_session": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/session.jsonl",
        "trace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/dual_shoes.md",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/workspace/.kinex/memos/dual_shoes.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian_move.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-02-20261113/workspace/.kinex/tools/cartesian_move.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 106,
        "observed_images": 23
      },
      "measured_protocol": {
        "seed": 1,
        "episodes": 1,
        "max_steps": 600,
        "max_control_steps": 600,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-progress-off-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 218,
        "published_frames": 218,
        "decode_ok": true,
        "duration_s": 5.45,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "fc0c22f735f349a930ab8069908a5d51d03e4007d62bf2df8bf8c9e7b38ac8eb",
        "source_sha256": {
          "third_person": "0be361e225ba4542dd698b1fd1bca313fcf1c90dd4c2425a7cadfda73228efa3",
          "left_wrist": "5919c31f7c6e1607a015d10ee659e1b4aa5a7f2bbb4a8a16144116e22f046489",
          "right_wrist": "66075d4eda0e105effd095b0c9748cca932b977b252ba7a6946955f91ac44846"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "1db67171ac55",
          "task": "robotwin/place-dual-shoes-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 218,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 218,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 21680000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 218,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 218,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 21680000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 218,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 218,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 21680000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-progress-off-20261113",
      "vla_model": "progVLA-jy",
      "vla_used": false,
      "superseded_by": "onboard-rerun-kinex-tcp-posestate-02-20261115",
      "archived_at": "2026-11-16T02:53:53.474121+00:00",
      "source_dataset_revision": "8cdd263ca04b3a249f3c8a4f66a653e2a890c8fb"
    },
    {
      "id": "native-full-kinex-tcp-posestate-05-20261113",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/05",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "05",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_superseded_by_onboard_rerun_20261115",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": true,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 36.24,
        "step_limit": 1200,
        "steps": 906,
        "success": true,
        "valid": true
      },
      "steps": 906,
      "simulation_time_s": 36.24,
      "wall_time_s": 2003.559551,
      "timing": {
        "simulation_time_s": 36.24,
        "wall_time_s": 2003.559551,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "native_instruction": "Grab the smooth bowl with glossy finish, stack them one by one.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9805112220368015,
        "cache_read_input_tokens": 1867264,
        "cache_reported_input_tokens": 1904378,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1904378,
        "cached_input_tokens": 1867264,
        "cost_usd": null,
        "fresh_input_tokens": 37114,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1904378,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1867264,
        "known_fresh_input_tokens": 37114,
        "known_input_tokens": 1904378,
        "known_output_tokens": 9719,
        "known_reasoning_output_tokens": 2409,
        "output_tokens": 9719,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 2409,
        "reasoning_reported_output_tokens": 9719,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 63,
          "cache_write_input_tokens": 63,
          "cache_write_reported_input_tokens": 63,
          "cached_input_tokens": 63,
          "fresh_input_tokens": 63,
          "input_tokens": 63,
          "output_tokens": 63,
          "reasoning_output_tokens": 63,
          "reasoning_reported_output_tokens": 63
        },
        "request_attempts": 63,
        "response_count": 63,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1914097,
        "uncached_input_tokens": 37114,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 9088,
            "input_tokens": 11821,
            "output_tokens": 14,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11856,
            "output_tokens": 27,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 11987,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12035,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12328,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12791,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12672,
            "input_tokens": 13832,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 13696,
            "input_tokens": 14646,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 14592,
            "input_tokens": 15463,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 15360,
            "input_tokens": 16289,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16128,
            "input_tokens": 17117,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17024,
            "input_tokens": 17909,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17792,
            "input_tokens": 18714,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 19518,
            "output_tokens": 30,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19456,
            "input_tokens": 19669,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 19584,
            "input_tokens": 21607,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21504,
            "input_tokens": 22177,
            "output_tokens": 98,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22144,
            "input_tokens": 22471,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22400,
            "input_tokens": 22662,
            "output_tokens": 77,
            "reasoning_output_tokens": 33
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22528,
            "input_tokens": 22949,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 24028,
            "output_tokens": 147,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 25230,
            "output_tokens": 284,
            "reasoning_output_tokens": 11
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 25909,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25728,
            "input_tokens": 26317,
            "output_tokens": 254,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26368,
            "input_tokens": 27197,
            "output_tokens": 167,
            "reasoning_output_tokens": 127
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27830,
            "output_tokens": 468,
            "reasoning_output_tokens": 68
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28160,
            "input_tokens": 28559,
            "output_tokens": 468,
            "reasoning_output_tokens": 58
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28928,
            "input_tokens": 29375,
            "output_tokens": 421,
            "reasoning_output_tokens": 341
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29696,
            "input_tokens": 30975,
            "output_tokens": 911,
            "reasoning_output_tokens": 81
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31744,
            "input_tokens": 31906,
            "output_tokens": 82,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32036,
            "output_tokens": 121,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32000,
            "input_tokens": 32673,
            "output_tokens": 45,
            "reasoning_output_tokens": 10
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 32740,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 32820,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33182,
            "output_tokens": 115,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33152,
            "input_tokens": 33680,
            "output_tokens": 257,
            "reasoning_output_tokens": 84
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 33997,
            "output_tokens": 52,
            "reasoning_output_tokens": 16
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33920,
            "input_tokens": 34071,
            "output_tokens": 48,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34048,
            "input_tokens": 34377,
            "output_tokens": 313,
            "reasoning_output_tokens": 187
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 35092,
            "output_tokens": 175,
            "reasoning_output_tokens": 47
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35200,
            "input_tokens": 35670,
            "output_tokens": 97,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36147,
            "output_tokens": 160,
            "reasoning_output_tokens": 73
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36224,
            "input_tokens": 37004,
            "output_tokens": 56,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36864,
            "input_tokens": 37082,
            "output_tokens": 100,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36992,
            "input_tokens": 37953,
            "output_tokens": 350,
            "reasoning_output_tokens": 94
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38144,
            "input_tokens": 38329,
            "output_tokens": 258,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 38892,
            "output_tokens": 452,
            "reasoning_output_tokens": 276
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39168,
            "input_tokens": 39944,
            "output_tokens": 129,
            "reasoning_output_tokens": 43
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39936,
            "input_tokens": 40364,
            "output_tokens": 569,
            "reasoning_output_tokens": 61
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40704,
            "input_tokens": 40955,
            "output_tokens": 81,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41233,
            "output_tokens": 236,
            "reasoning_output_tokens": 111
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 42207,
            "output_tokens": 34,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42112,
            "input_tokens": 42263,
            "output_tokens": 190,
            "reasoning_output_tokens": 67
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42240,
            "input_tokens": 43208,
            "output_tokens": 86,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43136,
            "input_tokens": 43356,
            "output_tokens": 36,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43590,
            "output_tokens": 141,
            "reasoning_output_tokens": 42
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 43757,
            "output_tokens": 133,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43648,
            "input_tokens": 44383,
            "output_tokens": 387,
            "reasoning_output_tokens": 301
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44672,
            "input_tokens": 45064,
            "output_tokens": 172,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45184,
            "input_tokens": 45378,
            "output_tokens": 68,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45312,
            "input_tokens": 45508,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 45775,
            "output_tokens": 682,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46336,
            "input_tokens": 46481,
            "output_tokens": 28,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 62,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 18,
          "bash": 1,
          "exec": 23,
          "devices": 5,
          "device": 1,
          "write": 3,
          "device_stop": 6,
          "edit": 3
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 906 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; full 50-task suite.",
        "execution": "Kinex + progVLA-jy 2026-11-13 result, replaced by the 2026-11-15 onboard rerun of tasks 01-05."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/transcript.json",
        "usage": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/usage.json",
        "analysis": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/analysis.json",
        "protocol": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/instruction.json",
        "verdict": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/episode.json",
        "media_check": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/evidence.zip",
        "timing": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/timing.json",
        "native_session": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/session.jsonl",
        "trace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack-bowls.md",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/workspace/.kinex/memos/stack-bowls.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/bowl_rims.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/workspace/.kinex/tools/bowl_rims.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/cartesian.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-05-20261113/workspace/.kinex/tools/cartesian.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 128,
        "observed_images": 14
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-progress-off-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 364,
        "published_frames": 364,
        "decode_ok": true,
        "duration_s": 9.1,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "b9cb073e8b07edfe6999eaa319bfada2a49abc8aee5560edd25626a639e0149f",
        "source_sha256": {
          "third_person": "88e437ee203859ec6dbf30aea95e8f2962bcb2c8317ff9981fe5553c313605ee",
          "left_wrist": "96b9a2be69ad695c3862fb0cc3bc12c11d66fa6eef30e4dd9e1dd3d81916698a",
          "right_wrist": "8a648f65329b2d67601a9dfe3546817e1a9999d68c2806e4735a8671abc9b7ac"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "8af2b2cc2a4f",
          "task": "robotwin/stack-bowls-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 364,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 364,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 36240000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 364,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 364,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 36240000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 364,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 364,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 36240000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-progress-off-20261113",
      "vla_model": "progVLA-jy",
      "vla_used": false,
      "superseded_by": "onboard-rerun-kinex-tcp-posestate-05-20261115",
      "archived_at": "2026-11-16T02:53:53.474121+00:00",
      "source_dataset_revision": "8cdd263ca04b3a249f3c8a4f66a653e2a890c8fb"
    },
    {
      "id": "native-full-kinex-tcp-posestate-06-20261113",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/06",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "06",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_superseded_by_onboard_rerun_20261116",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 34.36,
        "step_limit": 1200,
        "steps": 859,
        "success": true,
        "valid": true
      },
      "steps": 859,
      "simulation_time_s": 34.36,
      "wall_time_s": 1996.120292,
      "timing": {
        "simulation_time_s": 34.36,
        "wall_time_s": 1996.120292,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "native_instruction": "Move the blocks into a centered stack ordered red block, green block, blue block from bottom up.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.9718468126461567,
        "cache_read_input_tokens": 1760000,
        "cache_reported_input_tokens": 1810985,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 1810985,
        "cached_input_tokens": 1760000,
        "cost_usd": null,
        "fresh_input_tokens": 50985,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 1810985,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 1760000,
        "known_fresh_input_tokens": 50985,
        "known_input_tokens": 1810985,
        "known_output_tokens": 7295,
        "known_reasoning_output_tokens": 1196,
        "output_tokens": 7295,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 1196,
        "reasoning_reported_output_tokens": 7295,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 56,
          "cache_write_input_tokens": 56,
          "cache_write_reported_input_tokens": 56,
          "cached_input_tokens": 56,
          "fresh_input_tokens": 56,
          "input_tokens": 56,
          "output_tokens": 56,
          "reasoning_output_tokens": 56,
          "reasoning_reported_output_tokens": 56
        },
        "request_attempts": 56,
        "response_count": 56,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 1818280,
        "uncached_input_tokens": 50985,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11824,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11957,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12009,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12302,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12160,
            "input_tokens": 12511,
            "output_tokens": 194,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12544,
            "input_tokens": 16584,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 16512,
            "input_tokens": 17154,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17024,
            "input_tokens": 17406,
            "output_tokens": 85,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17408,
            "input_tokens": 18661,
            "output_tokens": 57,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 18825,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18688,
            "input_tokens": 19018,
            "output_tokens": 144,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18944,
            "input_tokens": 21431,
            "output_tokens": 38,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21248,
            "input_tokens": 21533,
            "output_tokens": 221,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21632,
            "input_tokens": 22149,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22016,
            "input_tokens": 22557,
            "output_tokens": 397,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22784,
            "input_tokens": 24023,
            "output_tokens": 231,
            "reasoning_output_tokens": 191
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24064,
            "input_tokens": 24720,
            "output_tokens": 1508,
            "reasoning_output_tokens": 168
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 26249,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 26112,
            "input_tokens": 27010,
            "output_tokens": 127,
            "reasoning_output_tokens": 8
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 27164,
            "output_tokens": 62,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27008,
            "input_tokens": 28117,
            "output_tokens": 123,
            "reasoning_output_tokens": 50
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 29138,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29056,
            "input_tokens": 29345,
            "output_tokens": 70,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29184,
            "input_tokens": 30312,
            "output_tokens": 103,
            "reasoning_output_tokens": 34
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30336,
            "input_tokens": 31301,
            "output_tokens": 108,
            "reasoning_output_tokens": 40
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31360,
            "input_tokens": 32295,
            "output_tokens": 55,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32256,
            "input_tokens": 32501,
            "output_tokens": 66,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32512,
            "input_tokens": 33459,
            "output_tokens": 130,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33408,
            "input_tokens": 33899,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33792,
            "input_tokens": 34864,
            "output_tokens": 56,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34816,
            "input_tokens": 35071,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34944,
            "input_tokens": 36033,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35968,
            "input_tokens": 36415,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36352,
            "input_tokens": 37379,
            "output_tokens": 54,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37248,
            "input_tokens": 37584,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 37504,
            "input_tokens": 38446,
            "output_tokens": 99,
            "reasoning_output_tokens": 28
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38400,
            "input_tokens": 39430,
            "output_tokens": 69,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39296,
            "input_tokens": 40386,
            "output_tokens": 171,
            "reasoning_output_tokens": 114
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 41013,
            "output_tokens": 68,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40960,
            "input_tokens": 41141,
            "output_tokens": 55,
            "reasoning_output_tokens": 18
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41088,
            "input_tokens": 41218,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41216,
            "input_tokens": 41634,
            "output_tokens": 106,
            "reasoning_output_tokens": 30
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41600,
            "input_tokens": 42634,
            "output_tokens": 73,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42624,
            "input_tokens": 43504,
            "output_tokens": 32,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43392,
            "input_tokens": 43598,
            "output_tokens": 85,
            "reasoning_output_tokens": 36
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43520,
            "input_tokens": 44051,
            "output_tokens": 272,
            "reasoning_output_tokens": 45
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44160,
            "input_tokens": 44350,
            "output_tokens": 115,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 44590,
            "output_tokens": 61,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44544,
            "input_tokens": 44802,
            "output_tokens": 72,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44800,
            "input_tokens": 45763,
            "output_tokens": 163,
            "reasoning_output_tokens": 87
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45824,
            "input_tokens": 46717,
            "output_tokens": 154,
            "reasoning_output_tokens": 77
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46720,
            "input_tokens": 47757,
            "output_tokens": 140,
            "reasoning_output_tokens": 64
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47744,
            "input_tokens": 48790,
            "output_tokens": 60,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48640,
            "input_tokens": 48992,
            "output_tokens": 72,
            "reasoning_output_tokens": 31
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49292,
            "output_tokens": 763,
            "reasoning_output_tokens": 90
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49920,
            "input_tokens": 50077,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 62,
        "model_tool_calls_by_name": {
          "read": 14,
          "ls": 1,
          "bash": 2,
          "exec": 33,
          "devices": 5,
          "device": 1,
          "write": 2,
          "edit": 2,
          "device_stop": 2
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 859 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; full 50-task suite.",
        "execution": "Kinex + progVLA-jy 2026-11-13 result, replaced by the 2026-11-16 onboard rerun of tasks 06-10."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/transcript.json",
        "usage": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/usage.json",
        "analysis": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/analysis.json",
        "protocol": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/instruction.json",
        "verdict": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/episode.json",
        "media_check": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/evidence.zip",
        "timing": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/timing.json",
        "native_session": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/session.jsonl",
        "trace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/stack_blocks.md",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/workspace/.kinex/memos/stack_blocks.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/stack_blocks.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-06-20261113/workspace/.kinex/tools/stack_blocks.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 121,
        "observed_images": 19
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 1200,
        "max_control_steps": 1200,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-progress-off-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 345,
        "published_frames": 345,
        "decode_ok": true,
        "duration_s": 8.625,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "3df523544e0aa1d2581c851908b1207aa7854a08b89223587fd3cccb26b92689",
        "source_sha256": {
          "third_person": "b650e6e338fc7d30c42d83182c33e14839e7a9f4b0fc06a76722708cf030104a",
          "left_wrist": "30cb0ee0c3ffda57dc558c9057c8d2e0b0fe25e6bd5b3913d15369d7cb6ead5f",
          "right_wrist": "ede2f8fe8e6c7a930b7e407ad5c621e5e16f379cd61270d4e73e30f7fe0c0e6b"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "e398cfd8f86c",
          "task": "robotwin/stack-blocks-three-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 345,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 345,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 34360000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 345,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 345,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 34360000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 345,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 345,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 34360000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-progress-off-20261113",
      "vla_model": "progVLA-jy",
      "vla_used": false,
      "superseded_by": "onboard-rerun-kinex-tcp-posestate-06-20261116",
      "archived_at": "2026-11-16T05:29:51.870521+00:00",
      "source_dataset_revision": "8fd3bb2f0051cb13bd6f7a48eb8a0ccab23c7217"
    },
    {
      "id": "native-full-kinex-tcp-posestate-07-20261113",
      "variant": "kinex-tcp-posestate",
      "task_key": "task03/07",
      "family": "task03",
      "source_family": "task13",
      "provider": "rosbridge",
      "slot": "07",
      "seed": 0,
      "source_episode_id": "episode_01",
      "status": "archived_superseded_by_onboard_rerun_20261116",
      "success": true,
      "valid": true,
      "protocol_valid": true,
      "execution": {
        "status": "finished",
        "reason": null
      },
      "verdict": {
        "episode": 1,
        "initialization": {
          "expert_plan_success": false,
          "requested_seed": 0
        },
        "native_reward": 1.0,
        "planned_episodes": 1,
        "reason": "success",
        "simulation_time_s": 33.4,
        "step_limit": 900,
        "steps": 835,
        "success": true,
        "valid": true
      },
      "steps": 835,
      "simulation_time_s": 33.4,
      "wall_time_s": 2407.409229,
      "timing": {
        "simulation_time_s": 33.4,
        "wall_time_s": 2407.409229,
        "wall_time_source": "Harbor agent_execution interval; initialization excluded"
      },
      "model": "GPT-6 Astra",
      "effort": "high",
      "harness": "kinex",
      "instruction": "Grab the cylindrical mug with handle on the table, rotate it, set it down in the middle, then hang it onto the rack with straight and slanted rods.",
      "native_instruction": "Grab the cylindrical mug with handle on the table, rotate it, set it down in the middle, then hang it onto the rack with straight and slanted rods.",
      "instruction_policy": "original_native",
      "usage": {
        "accounting": "reported-responses",
        "audit_complete": true,
        "audit_source": "Provider terminal response usage",
        "cache_hit_rate": 0.978414294131113,
        "cache_read_input_tokens": 2406272,
        "cache_reported_input_tokens": 2459359,
        "cache_write_basis": "provider-reported",
        "cache_write_input_tokens": 0,
        "cache_write_reported_input_tokens": 2459359,
        "cached_input_tokens": 2406272,
        "cost_usd": null,
        "fresh_input_tokens": 53087,
        "harbor_mismatches": {},
        "input_breakdown_complete": true,
        "input_tokens": 2459359,
        "known_cache_write_input_tokens": 0,
        "known_cached_input_tokens": 2406272,
        "known_fresh_input_tokens": 53087,
        "known_input_tokens": 2459359,
        "known_output_tokens": 14606,
        "known_reasoning_output_tokens": 7051,
        "output_tokens": 14606,
        "parser": "harbor/0.22.0",
        "reasoning_output_tokens": 7051,
        "reasoning_reported_output_tokens": 14606,
        "rejected_attempts": 0,
        "reported_responses": {
          "cache_reported_input_tokens": 66,
          "cache_write_input_tokens": 66,
          "cache_write_reported_input_tokens": 66,
          "cached_input_tokens": 66,
          "fresh_input_tokens": 66,
          "input_tokens": 66,
          "output_tokens": 66,
          "reasoning_output_tokens": 66,
          "reasoning_reported_output_tokens": 66
        },
        "request_attempts": 66,
        "response_count": 66,
        "response_ids_complete": true,
        "schema": "rlebench/token-usage/1",
        "source": "Harbor session metrics",
        "total_tokens": 2473965,
        "uncached_input_tokens": 53087,
        "unresolved_attempts": [],
        "usage_records": [
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 0,
            "input_tokens": 11834,
            "output_tokens": 19,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11648,
            "input_tokens": 11874,
            "output_tokens": 29,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11776,
            "input_tokens": 12007,
            "output_tokens": 20,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12055,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 11904,
            "input_tokens": 12348,
            "output_tokens": 253,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 12416,
            "input_tokens": 18054,
            "output_tokens": 39,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 17920,
            "input_tokens": 18624,
            "output_tokens": 49,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 18560,
            "input_tokens": 21827,
            "output_tokens": 78,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21760,
            "input_tokens": 22057,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 21888,
            "input_tokens": 22246,
            "output_tokens": 181,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 22272,
            "input_tokens": 24138,
            "output_tokens": 178,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 24192,
            "input_tokens": 25183,
            "output_tokens": 270,
            "reasoning_output_tokens": 21
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 25344,
            "input_tokens": 26001,
            "output_tokens": 1387,
            "reasoning_output_tokens": 499
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27264,
            "input_tokens": 27409,
            "output_tokens": 118,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 27392,
            "input_tokens": 28216,
            "output_tokens": 24,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28032,
            "input_tokens": 28467,
            "output_tokens": 40,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 28288,
            "input_tokens": 29464,
            "output_tokens": 58,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29312,
            "input_tokens": 29549,
            "output_tokens": 118,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 29568,
            "input_tokens": 30200,
            "output_tokens": 23,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30080,
            "input_tokens": 30285,
            "output_tokens": 53,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30365,
            "output_tokens": 35,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30208,
            "input_tokens": 30578,
            "output_tokens": 33,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30464,
            "input_tokens": 30718,
            "output_tokens": 146,
            "reasoning_output_tokens": 57
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 30720,
            "input_tokens": 31291,
            "output_tokens": 104,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31232,
            "input_tokens": 31714,
            "output_tokens": 229,
            "reasoning_output_tokens": 56
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 31872,
            "input_tokens": 32700,
            "output_tokens": 59,
            "reasoning_output_tokens": 17
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 32640,
            "input_tokens": 33657,
            "output_tokens": 42,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 33664,
            "input_tokens": 34280,
            "output_tokens": 131,
            "reasoning_output_tokens": 41
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34304,
            "input_tokens": 34438,
            "output_tokens": 168,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34666,
            "output_tokens": 25,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34560,
            "input_tokens": 34753,
            "output_tokens": 47,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 34688,
            "input_tokens": 35174,
            "output_tokens": 199,
            "reasoning_output_tokens": 71
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35328,
            "input_tokens": 35521,
            "output_tokens": 121,
            "reasoning_output_tokens": 26
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 35584,
            "input_tokens": 36071,
            "output_tokens": 132,
            "reasoning_output_tokens": 51
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 36096,
            "input_tokens": 37743,
            "output_tokens": 875,
            "reasoning_output_tokens": 405
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38528,
            "input_tokens": 38639,
            "output_tokens": 83,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 38656,
            "input_tokens": 39137,
            "output_tokens": 493,
            "reasoning_output_tokens": 362
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 39552,
            "input_tokens": 40429,
            "output_tokens": 125,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40448,
            "input_tokens": 40803,
            "output_tokens": 186,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 40832,
            "input_tokens": 41432,
            "output_tokens": 53,
            "reasoning_output_tokens": 14
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41344,
            "input_tokens": 41548,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 41472,
            "input_tokens": 41911,
            "output_tokens": 824,
            "reasoning_output_tokens": 738
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 42496,
            "input_tokens": 43281,
            "output_tokens": 134,
            "reasoning_output_tokens": 46
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 43264,
            "input_tokens": 43842,
            "output_tokens": 271,
            "reasoning_output_tokens": 208
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44032,
            "input_tokens": 44370,
            "output_tokens": 661,
            "reasoning_output_tokens": 486
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 44288,
            "input_tokens": 45577,
            "output_tokens": 223,
            "reasoning_output_tokens": 15
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 45440,
            "input_tokens": 46620,
            "output_tokens": 151,
            "reasoning_output_tokens": 55
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 46464,
            "input_tokens": 47198,
            "output_tokens": 159,
            "reasoning_output_tokens": 70
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 47232,
            "input_tokens": 47864,
            "output_tokens": 540,
            "reasoning_output_tokens": 400
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48256,
            "input_tokens": 48829,
            "output_tokens": 65,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48768,
            "input_tokens": 48916,
            "output_tokens": 111,
            "reasoning_output_tokens": 27
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 48896,
            "input_tokens": 49480,
            "output_tokens": 585,
            "reasoning_output_tokens": 378
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 49920,
            "input_tokens": 50349,
            "output_tokens": 431,
            "reasoning_output_tokens": 332
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 50688,
            "input_tokens": 51095,
            "output_tokens": 973,
            "reasoning_output_tokens": 896
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 51968,
            "input_tokens": 52132,
            "output_tokens": 108,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52096,
            "input_tokens": 52663,
            "output_tokens": 112,
            "reasoning_output_tokens": 32
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 52736,
            "input_tokens": 53239,
            "output_tokens": 545,
            "reasoning_output_tokens": 440
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 53632,
            "input_tokens": 54204,
            "output_tokens": 53,
            "reasoning_output_tokens": 19
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54144,
            "input_tokens": 54279,
            "output_tokens": 74,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 54272,
            "input_tokens": 54772,
            "output_tokens": 804,
            "reasoning_output_tokens": 698
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 55424,
            "input_tokens": 56000,
            "output_tokens": 277,
            "reasoning_output_tokens": 242
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56064,
            "input_tokens": 56299,
            "output_tokens": 106,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56192,
            "input_tokens": 56547,
            "output_tokens": 74,
            "reasoning_output_tokens": 22
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 56448,
            "input_tokens": 57016,
            "output_tokens": 316,
            "reasoning_output_tokens": 138
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57216,
            "input_tokens": 57359,
            "output_tokens": 640,
            "reasoning_output_tokens": 0
          },
          {
            "cache_write_input_tokens": 0,
            "cached_input_tokens": 57856,
            "input_tokens": 58022,
            "output_tokens": 37,
            "reasoning_output_tokens": 0
          }
        ],
        "usage_source": "provider-raw"
      },
      "call_activity": {
        "schema": "rlebench/call-activity/1",
        "source": "Kinex native trace",
        "model_tool_calls": 75,
        "model_tool_calls_by_name": {
          "ls": 2,
          "read": 18,
          "bash": 1,
          "exec": 34,
          "devices": 5,
          "device": 2,
          "write": 3,
          "edit": 4,
          "device_stop": 5,
          "relative_pose": 1
        },
        "python_instrumented_model_tool_calls": null,
        "python_tool_invocations": null,
        "nested_python_tool_invocations": null,
        "python_tool_invocations_by_origin": null,
        "python_device_rpc_attempts": null,
        "python_device_rpc_attempts_by_action": null,
        "python_device_rpc_errors": null,
        "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
      },
      "analysis": {
        "observed": "Native success after 835 rosbridge control frames; 0 optional VLA calls.",
        "instruction_alignment": "Original native unseen language from the pinned RoboTwin simulator.",
        "confidence": "One authored seed per task; full 50-task suite.",
        "execution": "Kinex + progVLA-jy 2026-11-13 result, replaced by the 2026-11-16 onboard rerun of tasks 06-10."
      },
      "links": {
        "video": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/presentation/video.mp4",
        "poster": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/presentation/poster.jpg",
        "terminal_frame": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/presentation/terminal.jpg",
        "transcript": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/transcript.json",
        "usage": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/usage.json",
        "analysis": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/analysis.json",
        "protocol": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/protocol.json",
        "native_goal": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/instruction.json",
        "verdict": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/episode.json",
        "media_check": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/presentation/media-validation.json",
        "recording_manifest": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/presentation/media-validation.json",
        "tool_calls": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/tool-calls.json",
        "artifact_index": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/artifact-index.json",
        "vla_io": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/vla-io.json",
        "evidence_archive": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/evidence.zip",
        "timing": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/timing.json",
        "native_session": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/session.jsonl",
        "trace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/trace.json",
        "provider_audit": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/provider-requests.jsonl",
        "provenance": "research/kinex-pose-hard-native-20261113/source-identities.json",
        "workspace": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/workspace.zip",
        "workspace_changes": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/workspace.json"
      },
      "resources": [
        {
          "name": ".kinex/memos/hanging-mug.md",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/workspace/.kinex/memos/hanging-mug.md",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/manipulate.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/workspace/.kinex/tools/manipulate.py",
          "kind": "Agent-authored episode resource"
        },
        {
          "name": ".kinex/tools/scene_geometry.py",
          "file": "experiments/kinex-pose-hard-native-20261113/native-full-kinex-tcp-posestate-07-20261113/workspace/.kinex/tools/scene_geometry.py",
          "kind": "Agent-authored episode resource"
        }
      ],
      "session_counts": {
        "visible_events": 144,
        "observed_images": 26
      },
      "measured_protocol": {
        "seed": 0,
        "episodes": 1,
        "max_steps": 900,
        "max_control_steps": 900,
        "task_config": "demo_randomized",
        "instruction_source": "native",
        "instruction_policy": "original_native",
        "provider": "rosbridge",
        "control_frequency_hz": 25,
        "mode": "stepped",
        "admission": "Authored task13 seed; no substitution, expert-success filter, or valid-outcome retry.",
        "runtime_revision": "pose-model-device-progress-off-20261113",
        "agent_protocol": "optional-model-device",
        "input_contract": "images-current-target-tcp-xyzw-v1",
        "progress_head": false,
        "source_action_period_s": 0.06,
        "action_chunk_size": 32,
        "ticks": 4,
        "tick_s": 0.04
      },
      "vla_calls": 0,
      "kinex_version": "1.1.0",
      "media": {
        "speed": 4,
        "source_fps": 10.0,
        "output_fps": 40.0,
        "width": 2880,
        "height": 720,
        "native_frames": 335,
        "published_frames": 335,
        "decode_ok": true,
        "duration_s": 8.375,
        "playback_kind": "simulation_time",
        "terminal_frame_retained": true,
        "camera_order": [
          "third_person",
          "left_wrist",
          "right_wrist"
        ],
        "sha256": "1611209d6b9099d3940678acdf8dba02da187b52c1ef3c9d28f0edde318574a2",
        "source_sha256": {
          "third_person": "433791a601af516ec12245c03a81bcb122d715b6083b6a1337da2a97805b0a50",
          "left_wrist": "f417c3b067a1ca833ba5846eb902f8af5c38d41ac44331770b103ec586b11c9c",
          "right_wrist": "50fe0d397ab71ba8fd03f16bee013ca58385e477dfc54bea2a87dd30dd1a2df0"
        },
        "recording": {
          "schema": "rosbridge/recording/1",
          "run_id": "58c35886618a",
          "task": "robotwin/hanging-mug-aloha-agilex-hard",
          "state": "complete",
          "views": {
            "third_person": {
              "received": 336,
              "dropped": 0,
              "segments": [
                {
                  "file": "third_person.mp4",
                  "frames": 335,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33400000000
                }
              ],
              "error": null
            },
            "left_wrist": {
              "received": 336,
              "dropped": 0,
              "segments": [
                {
                  "file": "left_wrist.mp4",
                  "frames": 335,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33400000000
                }
              ],
              "error": null
            },
            "right_wrist": {
              "received": 336,
              "dropped": 0,
              "segments": [
                {
                  "file": "right_wrist.mp4",
                  "frames": 335,
                  "width": 960,
                  "height": 720,
                  "fps": 10.0,
                  "first_stamp_ns": 0,
                  "last_stamp_ns": 33400000000
                }
              ],
              "error": null
            }
          },
          "type": "mp4",
          "spectator_views": [
            "third_person",
            "left_wrist",
            "right_wrist"
          ],
          "fps": 10.0,
          "topics": [],
          "error": null,
          "bag_messages": 0
        },
        "scope": "All recorded camera frames, composed side by side; 4x playback. No simulator replay."
      },
      "runtime_revision": "pose-model-device-progress-off-20261113",
      "vla_model": "progVLA-jy",
      "vla_used": false,
      "superseded_by": "onboard-rerun-kinex-tcp-posestate-07-20261116",
      "archived_at": "2026-11-16T05:29:51.870521+00:00",
      "source_dataset_revision": "8fd3bb2f0051cb13bd6f7a48eb8a0ccab23c7217"
    }
  ],
  "task_notes": {
    "task03/07": {
      "kind": "Native language initialization repair",
      "text": "For Kinex only, native mug/rack language bindings are retained when the expert path fails before finishing. The task, selected seed 0, instruction template, physics and verifier are unchanged. No completed model outcome was retried.",
      "evidence": "research/robotwin-hard-20261101/native-language-fix/receipt.json"
    }
  },
  "runtime_reviews": [],
  "variant_sources": {
    "standalone-vlact-clean": {
      "protocol": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/protocol.json",
      "source_identities": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/source-identities.json",
      "results": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/baseline/summary.json"
    },
    "kinex-tcp-posestate": {
      "protocol": "research/kinex-pose-hard-native-20261113/PROTOCOL.md",
      "source_identities": "research/kinex-pose-hard-native-20261113/source-identities.json",
      "results": "research/kinex-pose-hard-native-20261113/summary.json",
      "validation": "research/kinex-pose-hard-native-20261113/validation.json",
      "report": "research/kinex-pose-hard-native-20261113/REPORT.md"
    },
    "kinex-pose-base": {
      "protocol": "research/kinex-pose-hard-native-20261113/PROTOCOL.md",
      "source_identities": "research/kinex-pose-hard-native-20261113/source-identities.json",
      "results": "research/kinex-pose-hard-native-20261113/summary.json",
      "validation": "research/kinex-pose-hard-native-20261113/validation.json",
      "report": "research/kinex-pose-hard-native-20261113/REPORT.md"
    },
    "kinex-native": {
      "protocol": "research/kinex-hard-native-20261113/PROTOCOL.md",
      "source_identities": "research/kinex-hard-native-20261113/source-identities.json",
      "results": "research/kinex-hard-native-20261113/summary.json"
    },
    "kinex-vlact50k-native": {
      "protocol": "research/kinex-hard-native-20261113/PROTOCOL.md",
      "source_identities": "research/kinex-hard-native-20261113/source-identities.json",
      "results": "research/kinex-hard-native-20261113/summary.json"
    },
    "kinex-pose-film": {
      "protocol": "research/kinex-pose-film-hard-native-20261116/PROTOCOL.md",
      "report": "research/kinex-pose-film-hard-native-20261116/REPORT.md",
      "results": "research/kinex-pose-film-hard-native-20261116/summary.json",
      "source_identities": "research/kinex-pose-film-hard-native-20261116/source-identities.json",
      "validation": "research/kinex-pose-film-hard-native-20261116/validation.json"
    }
  },
  "budget_defaults": {},
  "instruction_protocol_notice": "Selected standalone runs use native unseen instructions. The VLAct OFT 50k baseline and all four conditions use identical frozen scenes. The earlier 16/50 baseline run, RLE history rows and interrupted baseline attempts are archived; raw evidence is preserved.",
  "standalone_native_campaign": {
    "planned": 250,
    "published": 250,
    "status": "complete",
    "running": 0,
    "pending": 0,
    "controller_updated_at": "2026-11-15T18:08:30.499777+00:00",
    "conditions": [
      "standalone-vlact-clean",
      "jy-mode1",
      "jy-mode2",
      "pose-base-mode1",
      "pose-base-mode2"
    ],
    "baseline": "VLAct OFT 50k baseline: 19/50.",
    "baseline_summary": {
      "variant": "standalone-vlact-clean",
      "successes": 19,
      "valid": 50,
      "summary": "https://huggingface.co/datasets/RLE-Bench/programmable-vla/resolve/d6dc53b7d2bb0f94eab9bc20b2b73de7dc600c8d/research/pose-standalone-native-hard-20261113/baseline/summary.json"
    }
  },
  "native_pose_progress": {
    "updated_at": "2026-11-13T22:53:29.954859+00:00",
    "state": "stopped_by_user",
    "planned": 100,
    "completed_valid": 11,
    "running": 0,
    "queued_attempts": 0,
    "frozen_pending": 79,
    "interrupted": 10,
    "automatic_resume": false,
    "infrastructure_attempts": 1,
    "engineering_scored": false
  },
  "native_pose_onboard_rerun": {
    "updated_at": "2026-11-16T02:53:53.474121+00:00",
    "variant": "kinex-tcp-posestate",
    "tasks": [
      "task03/01",
      "task03/02",
      "task03/03",
      "task03/04",
      "task03/05"
    ],
    "state": "complete",
    "completed_valid": 5,
    "successes": 2,
    "replaced_episodes": [
      "native-full-kinex-tcp-posestate-04-20261113",
      "native-full-kinex-tcp-posestate-01-20261113",
      "native-full-kinex-tcp-posestate-02-20261113",
      "native-full-kinex-tcp-posestate-05-20261113"
    ],
    "retained_earlier_episodes": [
      "native-full-kinex-tcp-posestate-06-20261113",
      "native-full-kinex-tcp-posestate-07-20261113"
    ],
    "archive_status": "archived_superseded_by_onboard_rerun_20261115",
    "protocol": "research/kinex-pose-hard-onboard-rerun-20261115/PROTOCOL.md"
  },
  "native_pose_onboard_rerun_06_10": {
    "updated_at": "2026-11-16T05:29:51.870521+00:00",
    "variant": "kinex-tcp-posestate",
    "tasks": [
      "task03/06",
      "task03/07",
      "task03/08",
      "task03/09",
      "task03/10"
    ],
    "state": "complete",
    "completed_valid": 5,
    "successes": 4,
    "replaced_episodes": [
      "native-full-kinex-tcp-posestate-06-20261113",
      "native-full-kinex-tcp-posestate-07-20261113"
    ],
    "retained_earlier_episodes": [
      "onboard-rerun-kinex-tcp-posestate-01-20261115",
      "onboard-rerun-kinex-tcp-posestate-02-20261115",
      "onboard-rerun-kinex-tcp-posestate-03-20261115",
      "onboard-rerun-kinex-tcp-posestate-04-20261115",
      "onboard-rerun-kinex-tcp-posestate-05-20261115"
    ],
    "archive_status": "archived_superseded_by_onboard_rerun_20261116",
    "protocol": "research/kinex-pose-hard-onboard-rerun-06-10-20261116/PROTOCOL.md"
  },
  "native_pose_hard50_20261116": {
    "updated_at": "2026-11-16T15:49:32.802420+00:00",
    "kind": "complete",
    "kinex-tcp-posestate": {
      "tasks": "11-50",
      "completed_valid": 40,
      "planned": 40,
      "running": 0,
      "successes": 23,
      "protocol": "research/kinex-pose-hard-onboard-11-50-20261116/PROTOCOL.md"
    },
    "kinex-pose-film": {
      "tasks": "01-50",
      "completed_valid": 50,
      "planned": 50,
      "running": 0,
      "successes": 31,
      "protocol": "research/kinex-pose-film-hard-native-20261116/PROTOCOL.md"
    },
    "engineering_scored": false,
    "valid_outcome_retries": 0
  }
}
